Model comparison
gpt-oss-20b vs Grok 3
Grok 3 is the stronger model overall, scoring 39.9 to 32.5 on the Noometry Index.
Last verified . 27 shared benchmarks.
Summary
- They share 27 benchmarks with published results for both. gpt-oss-20b scores higher in 2 categories and Grok 3 in 7 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in agentic & tool use, where Grok 3 leads 30.5 to 9.3.
- The biggest single-benchmark swing is IFEval: 73.2% for gpt-oss-20b and 88.4% for Grok 3.
- gpt-oss-20b has downloadable open weights; the other is API-only.
Side by side
| gpt-oss-20b | Grok 3 | |
|---|---|---|
| Provider | OpenAI | xAI |
| Noometry Index | 32.5 | 39.9 |
| Released | 2025-08-05 | 2025-04-09 |
| Weights | Open | Proprietary |
| Context window | 131K | — |
| Max output | 16K | — |
| Input $ / M tokens | $0.018 | — |
| Output $ / M tokens | $0.09 | — |
| Results tracked | 34 | 40 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Grok 3 leads
gpt-oss-20b: 37.6 (#192), Grok 3: 41.9 (#115)
| Benchmark | gpt-oss-20b | Grok 3 |
|---|---|---|
| WeirdML | 40.9% | 37.2% |
| LMArena Coding | 1306 | 1432 |
| Aider Polyglot | — | 53.3% |
| SciCode | 34.4% | — |
| ALE-Bench | 566.05 | — |
Agentic & Tool Use Grok 3 leads
gpt-oss-20b: 9.3 (#154), Grok 3: 30.5 (#76)
| Benchmark | gpt-oss-20b | Grok 3 |
|---|---|---|
| Terminal-Bench | 3.4% | — |
| BALROG | — | 29.5% |
Reasoning gpt-oss-20b leads
gpt-oss-20b: 19.3 (#261), Grok 3: 13.7 (#333)
| Benchmark | gpt-oss-20b | Grok 3 |
|---|---|---|
| Kagi LLM Benchmark | 53.2% | 61.3% |
| LMArena Hard Prompts | 1274 | 1434 |
| Epoch Capabilities Index | 137.82 | 138.33 |
| ARC-AGI-2 | — | 0% |
| SimpleBench | — | 36.1% |
| ARC-AGI-1 | — | 5.5% |
| CritPt | 1.4% | — |
| Chess Puzzles | 4% | — |
| DTBench | 68% | — |
| LMCA | 14.5% | — |
Math gpt-oss-20b leads
gpt-oss-20b: 39.4 (#103), Grok 3: 38.0 (#145)
| Benchmark | gpt-oss-20b | Grok 3 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 65.3% | 55.6% |
| Omni-MATH | 56.5% | 46.4% |
| LMArena Math | 1317 | 1391 |
| MATH Level 5 | — | 88.7% |
| FrontierMath (Feb 2025 set) | — | 3.8% |
| FrontierMath Tier 4 (v1) | — | 0% |
Knowledge Grok 3 leads
gpt-oss-20b: 34.6 (#195), Grok 3: 46.2 (#82)
| Benchmark | gpt-oss-20b | Grok 3 |
|---|---|---|
| GPQA Diamond | 60.8% | 75.8% |
| MMLU-Pro | 74% | 78.8% |
| GPQA (HELM) | 59.4% | 65% |
| LMArena Expert | 1258 | 1421 |
| Confabulations | — | 14.2% |
| Vectara Hallucination Rate | — | 5.8% |
Multilingual Grok 3 leads
gpt-oss-20b: 42.2 (#197), Grok 3: 52.3 (#87)
| Benchmark | gpt-oss-20b | Grok 3 |
|---|---|---|
| LMArena Non-English | 1268 | 1410 |
| LMArena Chinese | 1314 | 1448 |
| LMArena German | 1255 | 1431 |
| LMArena Japanese | 1244 | 1387 |
| LMArena Korean | 1236 | 1373 |
| LMArena Russian | 1278 | 1416 |
| LMArena Spanish | 1267 | 1417 |
| LMArena French | — | 1460 |
Instruction Following Grok 3 leads
gpt-oss-20b: 61.8 (#240), Grok 3: 75.0 (#73)
| Benchmark | gpt-oss-20b | Grok 3 |
|---|---|---|
| IFEval | 73.2% | 88.4% |
| LMArena Instruction Following | 1236 | 1409 |
Long Context Too close to call
gpt-oss-20b: 37.9 (#209), Grok 3: 38.7 (#192)
| Benchmark | gpt-oss-20b | Grok 3 |
|---|---|---|
| LMArena Longer Query | 1250 | 1439 |
| Fiction.LiveBench | — | 58.3% |
Writing & Preference Grok 3 leads
gpt-oss-20b: 35.5 (#265), Grok 3: 55.8 (#141)
| Benchmark | gpt-oss-20b | Grok 3 |
|---|---|---|
| LMArena Text | 1287 | 1426 |
| LMArena Creative Writing | 1201 | 1414 |
| EQ-Bench Creative Writing | 666 | 1186 |
| WildBench | 73.7% | 84.9% |
| LMArena Multi-Turn | 1268 | 1425 |
| Short-Story Creative Writing | — | 76.4% |
Frequently asked questions
Is gpt-oss-20b better than Grok 3?
Grok 3 is the stronger model overall, scoring 39.9 to 32.5 on the Noometry Index.
Is gpt-oss-20b or Grok 3 better for coding?
Grok 3 scores higher on coding benchmarks: 41.9 versus 37.6 in the Noometry coding category.
How many benchmarks do gpt-oss-20b and Grok 3 share?
27 benchmarks have published results for both models. gpt-oss-20b has 34 scored results on Noometry and Grok 3 has 40.