# Granite 4.2 3B Benchmark Scores & Performance

> Granite 4.2 3B by IBM has 19 source-displayable benchmark rows but no public overall score. It remains unranked.

Canonical page: https://benchlm.ai/models/granite-4-2-3b

Last updated: September 10, 2026

## Model Details

| Property | Value |
|----------|-------|
| Creator | IBM |
| Source Type | Open Weight |
| Reasoning Type | Reasoning |
| Context Window | 128K |
| Official model card | [IBM Granite 4.2 3B model card](https://huggingface.co/ibm-granite/granite-4.2-3b) |
| Overall Score | Coming soon |
| Overall Rank | Unranked |

## Family & Coverage

- Family: Granite 4.2
- Variant: 3b-dense (3B)
- Benchmarks covered: 19 of 434
- Sibling models: [Granite 4.2 8B](/models/granite-4-2-8b), [Granite 4.2 30B](/models/granite-4-2-30b)
- Coverage note: BenchLM currently has partial benchmark coverage for this model, so the overall score is conservative.

## Agentic Benchmarks

| Benchmark | Score |
|-----------|-------|
| [τ³-bench results](/benchmarks/tau3-bench) | 45.8% |
| [BFCL v4](/benchmarks/bfcl-v4) | 52.4% |
| [GDPval-AA](/benchmarks/gdpvalaanormalized) | 0.0% |

## Coding Benchmarks

| Benchmark | Score |
|-----------|-------|
| [LiveCodeBench v6](/benchmarks/livecodebench-v6) | 69.7% |
| [SciCode](/benchmarks/scicode) | 24.1% |
| [AA-SciCode](/benchmarks/aascicode) | 25.3% |

## Reasoning Benchmarks

| Benchmark | Score |
|-----------|-------|
| [AA-LCR](/benchmarks/lcr) | 24.3% |
| [CritPt](/benchmarks/critpt) | 0.0% |

## Knowledge Benchmarks

| Benchmark | Score |
|-----------|-------|
| [GPQA](/benchmarks/gpqa) | 54.8% |
| [MMLU-Pro](/benchmarks/mmlu-pro) | 67.8% |
| [Artificial Analysis Intelligence Index](/benchmarks/artificialanalysis) | 9.1% |
| [AA-GPQA Diamond](/benchmarks/aagpqadiamond) | 55.9% |
| [AA-HLE](/benchmarks/aahle) | 6.6% |
| [AA-Omniscience Index](/benchmarks/aaomniscienceindex) | -14.7% |
| [AA-Omniscience Accuracy](/benchmarks/omniscienceaccuracy) | 9.2% |
| [AA-Omniscience Hallucination Rate](/benchmarks/omnisciencehallucinationrate) | 26.3% |

## Instruction Following Benchmarks

| Benchmark | Score |
|-----------|-------|
| [IFBench](/benchmarks/ifbench) | 74.3% |

## Mathematics Benchmarks

| Benchmark | Score |
|-----------|-------|
| [AIME 2025](/benchmarks/aime2025) | 78.3% |
| [HMMT Feb 2025](/benchmarks/hmmtfeb2025) | 66.7% |

## Other IBM Models

- [Granite 4.2 8B](/models/granite-4-2-8b) - Score: 38.51
- [Granite-4.0-H-1B](/models/granite-4-0-h-1b) - Score: 37.9
- [Granite-4.0-1B](/models/granite-4-0-1b) - Score: 37.8
- [Granite-4.0-350M](/models/granite-4-0-350m) - Score: 37.7
- [Granite-4.0-H-350M](/models/granite-4-0-h-350m) - Score: 37.7
- [Granite 4.2 30B](/models/granite-4-2-30b) - Score: not computed
- [Granite Speech 5.0 470M TurboCTC](/models/granite-speech-5-0-470m-turboctc) - Score: not computed