GPT OSS 20B
| Released | 2025-08-05 |
|---|---|
| Price per 1M | $0.070 in · $0.30 out |
| Overall score | 40.6 · rank 187 |
| Context window | 131,072 tokens |
| Provider | OpenAI |
| Weights | Open weights |
Scores
| Dimension | Score | Rank | Benchmarks |
|---|---|---|---|
| Overall: Balanced | 40.6 | 187 | 11 |
| Overall: Coding | 44.3 | 161 | 6 |
| Overall: Reasoning | 43.6 | 170 | 5 |
| Overall: Agentic work | 38.0 | 199 | 6 |
| Overall: Frontend | 48.7 | 105 | 2 |
| Coding | 48.7 | 126 | 2 |
| Agents | 31.8 | 180 | 1 |
| Reasoning | 45.1 | 155 | 3 |
| Math | 47.7 | 133 | 1 |
| Knowledge | 33.2 | 154 | 1 |
| General | 31.3 | 181 | 3 |
| Value vs price | −1.0 | 133 |
Configurations
| Configuration | Balanced | Benchmarks |
|---|---|---|
| Unspecified setting | 41.1 ±2.4 | 8 |
| High | 43.8 ±4.1 | 4 |
| Low | 41.1 ±5.4 | 3 |
| Medium | 44.1 ±5.7 | 2 |
Evidence
Coding 48.7 ±3.5
| Benchmark | Result | Source | Date |
|---|---|---|---|
| SciCode | 38.9% | Epoch AI | 2026-11-11 |
| LiveCodeBench (Vals run) | 80.4% | Vals AI | 2026-09-01 |
Agents 31.8 ±2.5
| Benchmark | Result | Source | Date |
|---|---|---|---|
| Terminal-Bench 2.0 | 3.4% | Epoch AI | 2025-11-03 |
Reasoning 45.1 ±4.3
| Benchmark | Result | Source | Date |
|---|---|---|---|
| CritPt | 1.4% | Epoch AI | 2026-11-11 |
| Kagi Reasoning Benchmark | 53.2% | Kagi | 2026-11-11 |
| GPQA Diamond | 60.8% | Epoch AI | 2026-08-27 |
| GPQA Diamond (Vals run) | 68.9% | Vals AI | 2026-09-01 |
Math 47.7 ±3.6
| Benchmark | Result | Source | Date |
|---|---|---|---|
| OTIS Mock AIME 2024-2025 | 65.3% | Epoch AI | 2026-08-27 |
Knowledge 33.2 ±4.7
| Benchmark | Result | Source | Date |
|---|---|---|---|
| MMLU-Pro (Vals run) | 71.6% | Vals AI | 2026-09-01 |
General 31.3 ±4.7
| Benchmark | Result | Source | Date |
|---|---|---|---|
| LMArena Text | 1,318 | LMArena | 2026-11-09 |
| EQ-Bench Creative Writing v3 | 665 | EQ-Bench | 2026-11-11 |
| EQ-Bench Longform Creative Writing | 21.2% | EQ-Bench | 2026-11-11 |
Results not used in the score
| Benchmark | Result | Source | Why not used |
|---|---|---|---|
| CritPt | 0.0% | Epoch AI | A stronger configuration result was used |
| GPQA Diamond | 53.2% | Epoch AI | A stronger configuration result was used |
| GPQA Diamond | 46.0% | Epoch AI | A stronger configuration result was used |
| OTIS Mock AIME 2024-2025 | 40.3% | Epoch AI | A stronger configuration result was used |
| OTIS Mock AIME 2024-2025 | 50.8% | Epoch AI | A stronger configuration result was used |
| Terminal-Bench 2.0 | 3.1% | Epoch AI | A stronger configuration result was used |
Names matched to this model
gpt-oss-20b