Examenos

1237.2 · raw 1237.2 Elo

Headline · medium effort · Own vendor

Source/record date: 2026-10-19

Inception launch comparison; exact reported value 1237.2. Own-model vendor measurement. GDPval Elo; committee, reference pool and version not specified in the current figure. Do not map to GDPval-AA v2 or v2.1. All three named figure columns use medium. Execution date, seeds, task budgets and detailed harness are unpublished unless stated. Vendor claims are separate from evaluator-run results.

30.2% · raw 30.24 %

Headline · medium effort · Own vendor

Source/record date: 2026-10-19

Inception launch comparison; exact reported value 30.24. Own-model vendor measurement. Banking domain; simulator, harness and repeat aggregation are unpublished. All three named figure columns use medium. Execution date, seeds, task budgets and detailed harness are unpublished unless stated. Vendor claims are separate from evaluator-run results.

60.4% · raw 60.35 %

Headline · medium effort · Own vendor

Source/record date: 2026-10-19

Inception launch comparison; exact reported value 60.35. Own-model vendor measurement. Detailed scoring and harness are unpublished. All three named figure columns use medium. Execution date, seeds, task budgets and detailed harness are unpublished unless stated. Vendor claims are separate from evaluator-run results.

46.4% · raw 46.44 %

Headline · medium effort · Own vendor

Source/record date: 2026-10-19

Inception launch comparison; exact reported value 46.44. Own-model vendor measurement. Explicit public-set accuracy, not signed Omniscience Index or non-hallucination. All three named figure columns use medium. Execution date, seeds, task budgets and detailed harness are unpublished unless stated. Vendor claims are separate from evaluator-run results.

76.8% · raw 76.84 %

Headline · medium effort · Own vendor

Source/record date: 2026-10-19

Inception launch comparison; exact reported value 76.84. Own-model vendor measurement. Strict/loose and repeat aggregation not specified in this launch figure. All three named figure columns use medium. Execution date, seeds, task budgets and detailed harness are unpublished unless stated. Vendor claims are separate from evaluator-run results.

89.2% · raw 89.23 %

Headline · medium effort · Own vendor

Source/record date: 2026-10-19

Inception launch comparison; exact reported value 89.23. Own-model vendor measurement. Tool and repeat settings are unpublished. All three named figure columns use medium. Execution date, seeds, task budgets and detailed harness are unpublished unless stated. Vendor claims are separate from evaluator-run results.

77.7% · raw 77.67 %

Headline · medium effort · Own vendor

Source/record date: 2026-10-19

Inception launch comparison; exact reported value 77.67. Own-model vendor measurement. Explicit v1.1; not an unversioned legacy run. All three named figure columns use medium. Execution date, seeds, task budgets and detailed harness are unpublished unless stated. Vendor claims are separate from evaluator-run results.

75.50% · raw 75.5 %

Headline · medium effort · Own vendor

Source/record date: 2026-10-19

Inception launch comparison; exact reported value 75.5. Own-model vendor measurement. Needle count, token-length distribution and sample count are unpublished; not a known 8-needle/128K subset. All three named figure columns use medium. Execution date, seeds, task budgets and detailed harness are unpublished unless stated. Vendor claims are separate from evaluator-run results.

51.08% · raw 51.08 %

Headline · medium effort · Own vendor

Source/record date: 2026-10-19

Inception launch comparison; exact reported value 51.08. Own-model vendor measurement. Unspecified Atlas aggregate; not SWE-bench Verified or a named Refactoring/QnA/Test Writing split. All three named figure columns use medium. Execution date, seeds, task budgets and detailed harness are unpublished unless stated. Vendor claims are separate from evaluator-run results.

46.31% · raw 46.31 %

Headline · medium effort · Own vendor

Source/record date: 2026-10-19

Inception launch comparison; exact reported value 46.31. Own-model vendor measurement. Explicit v1; do not copy into v1.1 or Mini-SWE-V2. All three named figure columns use medium. Execution date, seeds, task budgets and detailed harness are unpublished unless stated. Vendor claims are separate from evaluator-run results.

55.18% · raw 55.18 %

Headline · medium effort · Own vendor

Source/record date: 2026-10-19

Inception launch comparison; exact reported value 55.18. Own-model vendor measurement. Explicit Plus variant; not BrowseComp or IFM Arabic evaluation. All three named figure columns use medium. Execution date, seeds, task budgets and detailed harness are unpublished unless stated. Vendor claims are separate from evaluator-run results.

71.3% · raw 71.33 %

Headline · medium effort · Own vendor

Source/record date: 2026-10-19

Inception launch comparison; exact reported value 71.33. Own-model vendor measurement. Search engine, budgets and aggregation are unpublished. All three named figure columns use medium. Execution date, seeds, task budgets and detailed harness are unpublished unless stated. Vendor claims are separate from evaluator-run results.

1068 tok/s · raw 1068 tok/s

Headline · medium effort · Own vendor

Source/record date: 2026-10-19

Inception launch comparison; exact reported value 1068. Own-model vendor measurement. Vendor NVIDIA claim, not an end-to-end latency guarantee. Peer ~130/~240 tok/s are approximations retained here as context rather than fabricated exact observations. All three named figure columns use medium. Execution date, seeds, task budgets and detailed harness are unpublished unless stated. Vendor claims are separate from evaluator-run results.

50.00% · raw 50 %

Headline · unknown effort · Own vendor

Source/record date: 2026-10-19

Inception launch comparison; exact reported value 50. Own-model vendor measurement. TinyFish production web-agent harness; 50 live navigation tasks, not the full Mind2Web leaderboard. Source says same token volume/steps as GPT-6.1 Sol but publishes no peer score. Exact harness/date unknown. Partner quotation; reasoning effort is unpublished. Execution date, seeds, task budgets and detailed harness are unpublished unless stated. Vendor claims are separate from evaluator-run results.

0.90 s · raw 0.9 s

Headline · unknown effort · Own vendor

Source/record date: 2026-10-19

Inception launch comparison; exact reported value 0.9. Own-model vendor measurement. Octen Sense API workload; median generation 0.9s. Task population, request lengths and sample count unpublished. Not a provider-wide throughput measurement. Partner quotation; reasoning effort is unpublished. Execution date, seeds, task budgets and detailed harness are unpublished unless stated. Vendor claims are separate from evaluator-run results.

2.20 s · raw 2.2 s

Headline · unknown effort · Own vendor

Source/record date: 2026-10-19

Inception launch comparison; exact reported value 2.2. Own-model vendor measurement. Octen Sense API workload; median end-to-end 2.2s. Search component about 80ms is not model time. Conditions/sample count unpublished. Partner quotation; reasoning effort is unpublished. Execution date, seeds, task budgets and detailed harness are unpublished unless stated. Vendor claims are separate from evaluator-run results.