IBMGranite 4.2 3B
Best configuration ranks #350 of 610 on the BenchLeader Index at 47.6 ±5.2. Last measured 2 Sept 2026. Released 25 Aug 2026.
- Blended price
- $0.053/M
- $0.030 in · $0.120 out
- Output speed
- 215 tok/s
- First answer
- 9.75 s
- Context
- 131k
- Overall index48
- Reasoning47
- Coding51
- Knowledge57
- Human preference46
- Long context37
- Composite40
Benchmark results
One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.
Reasoning
| Benchmark | default | Source |
|---|---|---|
| LMArena Hard Prompts | 1318#224 | LMArena |
| GPQA Diamond (AA)not in index | 55.9%#380 | Artificial Analysis |
| Humanity's Last Exam (AA)not in index | 6.6%#313 | Artificial Analysis |
Coding
| Benchmark | default | Source |
|---|---|---|
| LMArena Coding | 1380#192 | LMArena |
| SciCode (AA)not in index | 25.4%#153 | Artificial Analysis |
Knowledge
| Benchmark | default | Source |
|---|---|---|
| AA-Omniscience | -14.7#162 | Artificial Analysis |
Human preference
| Benchmark | default | Source |
|---|---|---|
| LMArena Text | 1296#234 | LMArena |
Long context
| Benchmark | default | Source |
|---|---|---|
| AA-LCR | 24.3%#343 | Artificial Analysis |
Composite
| Benchmark | default | Source |
|---|---|---|
| AA Intelligence Index | 9.1#337 | Artificial Analysis |
What a task costs
Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0000 | – | 11.1 s |
| Summarise a 30-page report | 12,000 / 600 | $0.0004 | – | 12.5 s |
| Code edit | 6,000 / 1,500 | $0.0004 | – | 16.7 s |
| Agentic coding session | 60,000 / 4,000 | $0.0023 | – | 28.3 s |
| Structured extraction | 2,000 / 200 | $0.0001 | – | 10.7 s |
See also
Data as of 9 Sept 2026. Compare with another model.