Examenos

Correct answers over all 1,934 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 1934

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Equal-weight mean over the 41 categories.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 1934

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Accuracy on the 592 questions new in v2.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 592

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Accuracy on the 192 logical additions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 192

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Accuracy on the 400 practical additions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 400

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Accuracy on the frozen original inputs inside the v2 bank.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 1342

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Pooled primary complete-response client time, network included.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 1934

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Pooled primary complete-response client time, network included.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 1934

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Lookup and filtering: 43 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 43

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Forward conditionals: 46 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 46

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Necessary conditions: 52 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 52

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Boolean scope: 54 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 54

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Quantifiers and sets: 49 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 49

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Comparisons: 40 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 40

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Scheduling: 76 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 76

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

One-to-one assignments: 67 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 67

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Exact truth counting: 71 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 71

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Truth-teller consistency: 74 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 74

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Sufficiency and inconsistency: 41 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 41

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Object swaps: 52 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 52

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Sequential procedures: 48 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 48

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Temporal intervals: 42 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 42

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Spatial tracking: 49 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 49

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Reference binding: 50 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 50

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Beliefs and perspectives: 50 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 50

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Causal intervention: 50 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 50

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Exceptions and priorities: 58 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 58

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Planning: 43 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 43

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Directed reachability: 43 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 43

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Preference ordering: 51 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 51

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Rule counterexamples: 52 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 52

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Authority and stale records: 43 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 43

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Fault diagnosis: 50 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 50

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Probabilistic risk: 40 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 40

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Value of information: 40 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 40

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Robust decisions without a prior: 40 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 40

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Strategic decisions: 40 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 40

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Concurrent-history consistency: 40 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 40

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Information-flow security: 40 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 40

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Understanding requests and routing: 40 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 40

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Recognizing multiple requested actions: 40 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 40

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Tracking current conversational intention: 40 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 40

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Selecting a tool or requesting clarification: 40 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 40

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Checking authorization and scope: 40 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 40

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Making decisions with untrusted content: 40 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 40

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Applying a contextual content policy: 40 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 40

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Extracting and normalizing facts: 40 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 40

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Checking claims against supplied evidence: 40 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 40

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Assigning severity or priority from a rubric: 40 questions.

Celeris 1.1 Decision; Hosted native; returned identities {"celeris-1.1-decision": 1934}; collected 2026-10-25 to 2026-11-01 America/Toronto; one primary response per input, no scored retries.

Exact aggregate from the release statistics.

Development benchmark, not an independently sampled test: 1,934 questions in 41 categories, the frozen 1,342-question bank plus 192 logical and 400 practical additions. The bank was built with model-guided development, and Jev plus retained Microsoft and Drex feedback are development exposure. Related variants and a single recorded primary response per input limit generalization. The scoring key stays private, so public checks cover coverage, timing and aggregate consistency but cannot rescore answers. Strict accuracy counts a wrong usable answer and an unsuccessful response in the same fixed denominator. Timing pools primary client times with the network included; providers, precision, concurrency, routes and dates differ, retained and new timings can span sessions, and quick failures can reduce all-primary latency. The source published a separate cost ledger on 1 November 2026, stored as decision-reference-v2-cost-per-1000, which supersedes its earlier “cost not consolidated” note. Older 1,342-only scores are not converted into this bank. Not ranked here.

Checked / reported: 2026-11-02 · Samples: 40

Snapshot: 2026-11-01 · Revision: f5fa9cfafccd97fcfbdb0336e14284faaa8d2d35 · Published summary observations; upstream task/result terms retained

Captured-tariff estimate for primary calls without a reported charge, in USD per 1,000 questions.

Celeris 1.1 Decision; System 1 v2 route; evidence captured_tariff_or_mixed.

1934 calls at a captured tariff, 0 on free access, 0 unknown, out of 1,934.

Tariff-based estimate from the same ledger, kept apart from the reported cost dataset. These calls have no provider-reported charge: the ledger applied a captured tariff, and calls on free access during a launch week add nothing, so a value can understate the paid rate. Providers and routes differ; these are not current tariffs. Development benchmark; not ranked here.

Checked / reported: 2026-11-02 · Samples: 1934

Snapshot: 2026-11-02 · Revision: 5b47806ba527db5e1abc88578789f3eae6e8e646 · Published summary observations; upstream task/result terms retained