# Claude 3 Haiku vs gpt-oss-20b

> gpt-oss-20b is the stronger model overall, scoring 32.5 to 25.9 on the Noometry Index.

- Canonical page: https://noometry.com/compare/claude-3-haiku-vs-gpt-oss-20b
- Last updated: 2026-10-12
- Shared benchmarks: 24

## Summary

- They share 24 benchmarks with published results for both. Claude 3 Haiku scores higher in 0 categories and gpt-oss-20b in 8 categories; 7 gaps are clear of the uncertainty.
- The widest gap is in math, where gpt-oss-20b leads 39.4 to 9.8.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 1.8% for Claude 3 Haiku and 65.3% for gpt-oss-20b.
- gpt-oss-20b has downloadable open weights; the other is API-only.

## Snapshot

| | Claude 3 Haiku | gpt-oss-20b |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 25.9 | 32.5 |
| Rank | 340 | 255 |
| Context | — | 131K |
| Input $/M | — | $0.018 |
| Output $/M | — | $0.09 |
| Weights | Proprietary | Open |

## Coding

- Claude 3 Haiku: 26.4 (#325)
- gpt-oss-20b: 37.6 (#192)

| Benchmark | Claude 3 Haiku | gpt-oss-20b |
|---|---|---|
| WeirdML | 9.8% | 40.9% |
| LMArena Coding | 1199 | 1306 |
| SciCode | — | 34.4% |
| BigCodeBench Instruct | 39.4% | — |
| BigCodeBench Complete | 50.1% | — |
| CadEval | 12% | — |
| ALE-Bench | — | 566.05 |
| HumanEval+ | 68.9% | — |
| MBPP+ | 68.8% | — |

## Agentic & Tool Use

- Claude 3 Haiku: —
- gpt-oss-20b: 9.3 (#154)

| Benchmark | Claude 3 Haiku | gpt-oss-20b |
|---|---|---|
| Terminal-Bench | — | 3.4% |

## Reasoning

- Claude 3 Haiku: 16.3 (#307)
- gpt-oss-20b: 19.3 (#261)

| Benchmark | Claude 3 Haiku | gpt-oss-20b |
|---|---|---|
| Kagi LLM Benchmark | 34.2% | 53.2% |
| LMArena Hard Prompts | 1174 | 1274 |
| DTBench | 50.1% | 68% |
| LMCA | 8.8% | 14.5% |
| Epoch Capabilities Index | 118.35 | 137.82 |
| CritPt | — | 1.4% |
| Chess Puzzles | — | 4% |
| ForecastBench | 53.2 | — |
| WinoGrande | 74.2% | — |

## Math

- Claude 3 Haiku: 9.8 (#319)
- gpt-oss-20b: 39.4 (#103)

| Benchmark | Claude 3 Haiku | gpt-oss-20b |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 1.8% | 65.3% |
| LMArena Math | 1188 | 1317 |
| Omni-MATH | — | 56.5% |
| MATH Level 5 | 14.9% | — |

## Knowledge

- Claude 3 Haiku: 17.3 (#285)
- gpt-oss-20b: 34.6 (#195)

| Benchmark | Claude 3 Haiku | gpt-oss-20b |
|---|---|---|
| GPQA Diamond | 36.3% | 60.8% |
| LMArena Expert | 1148 | 1258 |
| MMLU-Pro | — | 74% |
| Confabulations | 34.2% | — |
| GPQA (HELM) | — | 59.4% |
| MMLU | 73.8% | — |

## Multimodal

- Claude 3 Haiku: 23.6 (#128)
- gpt-oss-20b: —

| Benchmark | Claude 3 Haiku | gpt-oss-20b |
|---|---|---|
| LMArena Vision | 950 | — |
| ScienceQA | 72% | — |

## Multilingual

- Claude 3 Haiku: 36.0 (#243)
- gpt-oss-20b: 42.2 (#197)

| Benchmark | Claude 3 Haiku | gpt-oss-20b |
|---|---|---|
| LMArena Non-English | 1178 | 1268 |
| LMArena Chinese | 1155 | 1314 |
| LMArena German | 1174 | 1255 |
| LMArena Japanese | 1102 | 1244 |
| LMArena Korean | 1109 | 1236 |
| LMArena Russian | 1204 | 1278 |
| LMArena Spanish | 1166 | 1267 |
| LMArena French | 1195 | — |

## Instruction Following

- Claude 3 Haiku: 61.3 (#247)
- gpt-oss-20b: 61.8 (#240)

| Benchmark | Claude 3 Haiku | gpt-oss-20b |
|---|---|---|
| LMArena Instruction Following | 1173 | 1236 |
| IFEval | — | 73.2% |

## Long Context

- Claude 3 Haiku: 36.1 (#237)
- gpt-oss-20b: 37.9 (#209)

| Benchmark | Claude 3 Haiku | gpt-oss-20b |
|---|---|---|
| LMArena Longer Query | 1190 | 1250 |

## Writing & Preference

- Claude 3 Haiku: 29.7 (#291)
- gpt-oss-20b: 35.5 (#265)

| Benchmark | Claude 3 Haiku | gpt-oss-20b |
|---|---|---|
| LMArena Text | 1195 | 1287 |
| LMArena Creative Writing | 1157 | 1201 |
| EQ-Bench Creative Writing | 717 | 666 |
| LMArena Multi-Turn | 1190 | 1268 |
| WildBench | — | 73.7% |

## FAQ

### Is Claude 3 Haiku better than gpt-oss-20b?

gpt-oss-20b is the stronger model overall, scoring 32.5 to 25.9 on the Noometry Index.

### Is Claude 3 Haiku or gpt-oss-20b better for coding?

gpt-oss-20b scores higher on coding benchmarks: 37.6 versus 26.4 in the Noometry coding category.

### How many benchmarks do Claude 3 Haiku and gpt-oss-20b share?

24 benchmarks have published results for both models. Claude 3 Haiku has 37 scored results on Noometry and gpt-oss-20b has 34.