Granite Switch 4.1.1 30B · Decision
Explore Granite Switch 4.1.1 30B · Decision: exact decision configuration, published specifications, source-linked benchmarks and API tariffs when available.
← All decision models · Compare this decision system →
6 recorded observations. Published results retain their source and setup; no Examenos evaluation was run. Missing results are not zero. How to read these comparisons →
Official / vendor · IBM (vendor measured)
| Evaluation | Result | Evidence |
|---|---|---|
| IBM Granite Switch 4.1.1 cards · Decision Index v0.3 public overall | 51.33 | Setup & sourceDecision Index v0.3 public set, chance-corrected weighted aggregate, as reported by IBM on the Granite Switch 4.1.1 cards Granite Switch 4.1.1 30B decision aLoRA, card revision eb206725d03c0f1c409ce3aa512afe325fdf8731. From the Granite Switch 4.1.1 card Decision Index table. IBM vendor measurement of the revision 4.1.1 decision aLoRA on the public Jev Decision Index v0.3 set with the documented invocation recipe (fixed system prompt, State turn on the frozen base, ### DECISION turn, first assistant token read as a softmax over option-label logits). Scores ×100; Overall is the chance-corrected weighted aggregate. The card does not name the area metric, so area values are kept as printed points. Not comparable with community Decision Index 0.4/0.4.1 Full or Public. The 4.1 cards scored a different 1,068-row fast subset, which is not stored. Checked / reported: 2026-10-29 · Samples: Not published Snapshot: 2026-10-27 · Revision: 8B fb4aa7877e13e5fe09ca328713dc1db72a3f93ce; 30B eb206725d03c0f1c409ce3aa512afe325fdf8731 · Published observations; result reuse terms not specified Result source · Benchmark source |
| IBM Granite Switch 4.1.1 cards · Decision Index v0.3 public knowledge | 34.8 points | Setup & sourceDecision Index v0.3 public set, Knowledge area, as reported by IBM on the Granite Switch 4.1.1 cards Granite Switch 4.1.1 30B decision aLoRA, card revision eb206725d03c0f1c409ce3aa512afe325fdf8731. From the Granite Switch 4.1.1 card Decision Index table. IBM vendor measurement of the revision 4.1.1 decision aLoRA on the public Jev Decision Index v0.3 set with the documented invocation recipe (fixed system prompt, State turn on the frozen base, ### DECISION turn, first assistant token read as a softmax over option-label logits). Scores ×100; Overall is the chance-corrected weighted aggregate. The card does not name the area metric, so area values are kept as printed points. Not comparable with community Decision Index 0.4/0.4.1 Full or Public. The 4.1 cards scored a different 1,068-row fast subset, which is not stored. Checked / reported: 2026-10-29 · Samples: Not published Snapshot: 2026-10-27 · Revision: 8B fb4aa7877e13e5fe09ca328713dc1db72a3f93ce; 30B eb206725d03c0f1c409ce3aa512afe325fdf8731 · Published observations; result reuse terms not specified Result source · Benchmark source |
| IBM Granite Switch 4.1.1 cards · Decision Index v0.3 public language | 58.5 points | Setup & sourceDecision Index v0.3 public set, Language area, as reported by IBM on the Granite Switch 4.1.1 cards Granite Switch 4.1.1 30B decision aLoRA, card revision eb206725d03c0f1c409ce3aa512afe325fdf8731. From the Granite Switch 4.1.1 card Decision Index table. IBM vendor measurement of the revision 4.1.1 decision aLoRA on the public Jev Decision Index v0.3 set with the documented invocation recipe (fixed system prompt, State turn on the frozen base, ### DECISION turn, first assistant token read as a softmax over option-label logits). Scores ×100; Overall is the chance-corrected weighted aggregate. The card does not name the area metric, so area values are kept as printed points. Not comparable with community Decision Index 0.4/0.4.1 Full or Public. The 4.1 cards scored a different 1,068-row fast subset, which is not stored. Checked / reported: 2026-10-29 · Samples: Not published Snapshot: 2026-10-27 · Revision: 8B fb4aa7877e13e5fe09ca328713dc1db72a3f93ce; 30B eb206725d03c0f1c409ce3aa512afe325fdf8731 · Published observations; result reuse terms not specified Result source · Benchmark source |
| IBM Granite Switch 4.1.1 cards · Decision Index v0.3 public retrieval | 59.8 points | Setup & sourceDecision Index v0.3 public set, Retrieval area, as reported by IBM on the Granite Switch 4.1.1 cards Granite Switch 4.1.1 30B decision aLoRA, card revision eb206725d03c0f1c409ce3aa512afe325fdf8731. From the Granite Switch 4.1.1 card Decision Index table. IBM vendor measurement of the revision 4.1.1 decision aLoRA on the public Jev Decision Index v0.3 set with the documented invocation recipe (fixed system prompt, State turn on the frozen base, ### DECISION turn, first assistant token read as a softmax over option-label logits). Scores ×100; Overall is the chance-corrected weighted aggregate. The card does not name the area metric, so area values are kept as printed points. Not comparable with community Decision Index 0.4/0.4.1 Full or Public. The 4.1 cards scored a different 1,068-row fast subset, which is not stored. Checked / reported: 2026-10-29 · Samples: Not published Snapshot: 2026-10-27 · Revision: 8B fb4aa7877e13e5fe09ca328713dc1db72a3f93ce; 30B eb206725d03c0f1c409ce3aa512afe325fdf8731 · Published observations; result reuse terms not specified Result source · Benchmark source |
| IBM Granite Switch 4.1.1 cards · Decision Index v0.3 public tools | 69.0 points | Setup & sourceDecision Index v0.3 public set, Tools area, as reported by IBM on the Granite Switch 4.1.1 cards Granite Switch 4.1.1 30B decision aLoRA, card revision eb206725d03c0f1c409ce3aa512afe325fdf8731. From the Granite Switch 4.1.1 card Decision Index table. IBM vendor measurement of the revision 4.1.1 decision aLoRA on the public Jev Decision Index v0.3 set with the documented invocation recipe (fixed system prompt, State turn on the frozen base, ### DECISION turn, first assistant token read as a softmax over option-label logits). Scores ×100; Overall is the chance-corrected weighted aggregate. The card does not name the area metric, so area values are kept as printed points. Not comparable with community Decision Index 0.4/0.4.1 Full or Public. The 4.1 cards scored a different 1,068-row fast subset, which is not stored. Checked / reported: 2026-10-29 · Samples: Not published Snapshot: 2026-10-27 · Revision: 8B fb4aa7877e13e5fe09ca328713dc1db72a3f93ce; 30B eb206725d03c0f1c409ce3aa512afe325fdf8731 · Published observations; result reuse terms not specified Result source · Benchmark source |
| IBM Granite Switch 4.1.1 cards · Decision Index v0.3 public arts | 25.9 points | Setup & sourceDecision Index v0.3 public set, Arts area, as reported by IBM on the Granite Switch 4.1.1 cards Granite Switch 4.1.1 30B decision aLoRA, card revision eb206725d03c0f1c409ce3aa512afe325fdf8731. From the Granite Switch 4.1.1 card Decision Index table. IBM vendor measurement of the revision 4.1.1 decision aLoRA on the public Jev Decision Index v0.3 set with the documented invocation recipe (fixed system prompt, State turn on the frozen base, ### DECISION turn, first assistant token read as a softmax over option-label logits). Scores ×100; Overall is the chance-corrected weighted aggregate. The card does not name the area metric, so area values are kept as printed points. Not comparable with community Decision Index 0.4/0.4.1 Full or Public. The 4.1 cards scored a different 1,068-row fast subset, which is not stored. Checked / reported: 2026-10-29 · Samples: Not published Snapshot: 2026-10-27 · Revision: 8B fb4aa7877e13e5fe09ca328713dc1db72a3f93ce; 30B eb206725d03c0f1c409ce3aa512afe325fdf8731 · Published observations; result reuse terms not specified Result source · Benchmark source |