# gpt-oss-20b vs Qwen3 8B

> Qwen3 8B is the stronger model overall, scoring 33.7 to 32.5 on the Noometry Index. gpt-oss-20b costs 8.6× less per token, which makes it the better buy when Qwen3 8B's lead doesn't matter for your workload.

- Canonical page: https://noometry.com/compare/gpt-oss-20b-vs-qwen3-8b
- Last updated: 2026-10-12
- Shared benchmarks: 8

## Summary

- They share 8 benchmarks with published results for both. gpt-oss-20b scores higher in 4 categories and Qwen3 8B in 2 categories; 5 gaps are clear of the uncertainty.
- The widest gap is in agentic & tool use, where Qwen3 8B leads 30.2 to 9.3.
- The biggest single-benchmark swing is SciCode: 34.4% for gpt-oss-20b and 22.6% for Qwen3 8B.
- gpt-oss-20b is cheaper at $0.018 / $0.09 per million input/output tokens, against $0.18 / $0.70 for Qwen3 8B.

## Snapshot

| | gpt-oss-20b | Qwen3 8B |
|---|---|---|
| Provider | OpenAI | Alibaba (Qwen) |
| Noometry Index | 32.5 | 33.7 |
| Rank | 255 | 238 |
| Context | 131K | 131K |
| Input $/M | $0.018 | $0.18 |
| Output $/M | $0.09 | $0.70 |
| Weights | Open | Open |

## Coding

- gpt-oss-20b: 37.6 (#192)
- Qwen3 8B: 34.0 (#248)

| Benchmark | gpt-oss-20b | Qwen3 8B |
|---|---|---|
| SciCode | 34.4% | 22.6% |
| WeirdML | 40.9% | — |
| LMArena Coding | 1306 | — |
| ALE-Bench | 566.05 | — |

## Agentic & Tool Use

- gpt-oss-20b: 9.3 (#154)
- Qwen3 8B: 30.2 (#78)

| Benchmark | gpt-oss-20b | Qwen3 8B |
|---|---|---|
| Terminal-Bench | 3.4% | — |
| Berkeley Function Calling Leaderboard | — | 42.6% |

## Reasoning

- gpt-oss-20b: 19.3 (#261)
- Qwen3 8B: 16.6 (#303)

| Benchmark | gpt-oss-20b | Qwen3 8B |
|---|---|---|
| CritPt | 1.4% | 0% |
| Chess Puzzles | 4% | 5% |
| DTBench | 68% | 59.7% |
| LMCA | 14.5% | 8.8% |
| Epoch Capabilities Index | 137.82 | 136.17 |
| Kagi LLM Benchmark | 53.2% | — |
| LMArena Hard Prompts | 1274 | — |

## Math

- gpt-oss-20b: 39.4 (#103)
- Qwen3 8B: 34.9 (#191)

| Benchmark | gpt-oss-20b | Qwen3 8B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 65.3% | 56.1% |
| Omni-MATH | 56.5% | — |
| LMArena Math | 1317 | — |

## Knowledge

- gpt-oss-20b: 34.6 (#195)
- Qwen3 8B: 36.1 (#173)

| Benchmark | gpt-oss-20b | Qwen3 8B |
|---|---|---|
| GPQA Diamond | 60.8% | 56.8% |
| MMLU-Pro | 74% | — |
| Vectara Hallucination Rate | — | 4.8% |
| GPQA (HELM) | 59.4% | — |
| LMArena Expert | 1258 | — |

## Multilingual

- gpt-oss-20b: 42.2 (#197)
- Qwen3 8B: —

| Benchmark | gpt-oss-20b | Qwen3 8B |
|---|---|---|
| LMArena Non-English | 1268 | — |
| LMArena Chinese | 1314 | — |
| LMArena German | 1255 | — |
| LMArena Japanese | 1244 | — |
| LMArena Korean | 1236 | — |
| LMArena Russian | 1278 | — |
| LMArena Spanish | 1267 | — |

## Instruction Following

- gpt-oss-20b: 61.8 (#240)
- Qwen3 8B: —

| Benchmark | gpt-oss-20b | Qwen3 8B |
|---|---|---|
| IFEval | 73.2% | — |
| LMArena Instruction Following | 1236 | — |

## Long Context

- gpt-oss-20b: 37.9 (#209)
- Qwen3 8B: 37.9 (#210)

| Benchmark | gpt-oss-20b | Qwen3 8B |
|---|---|---|
| Fiction.LiveBench | — | 62.1% |
| LMArena Longer Query | 1250 | — |

## Writing & Preference

- gpt-oss-20b: 35.5 (#265)
- Qwen3 8B: —

| Benchmark | gpt-oss-20b | Qwen3 8B |
|---|---|---|
| LMArena Text | 1287 | — |
| LMArena Creative Writing | 1201 | — |
| EQ-Bench Creative Writing | 666 | — |
| WildBench | 73.7% | — |
| LMArena Multi-Turn | 1268 | — |

## FAQ

### Is gpt-oss-20b better than Qwen3 8B?

Qwen3 8B is the stronger model overall, scoring 33.7 to 32.5 on the Noometry Index. gpt-oss-20b costs 8.6× less per token, which makes it the better buy when Qwen3 8B's lead doesn't matter for your workload.

### Which is cheaper, gpt-oss-20b or Qwen3 8B?

gpt-oss-20b is cheaper. It lists at $0.018 per million input tokens and $0.09 per million output tokens; Qwen3 8B lists at $0.18 and $0.70.

### Is gpt-oss-20b or Qwen3 8B better for coding?

gpt-oss-20b scores higher on coding benchmarks: 37.6 versus 34.0 in the Noometry coding category.

### Which has the bigger context window?

Both accept 131K tokens.

### How many benchmarks do gpt-oss-20b and Qwen3 8B share?

8 benchmarks have published results for both models. gpt-oss-20b has 34 scored results on Noometry and Qwen3 8B has 11.