Model comparison
DeepSeek-V3 vs gpt-oss-120b
DeepSeek-V3 is the stronger model overall, scoring 39.2 to 36.0 on the Noometry Index. gpt-oss-120b costs 5.8× less per token, which makes it the better buy when DeepSeek-V3's lead doesn't matter for your workload.
Last verified . 39 shared benchmarks.
DeepSeekSummary
- They share 39 benchmarks with published results for both. DeepSeek-V3 scores higher in 6 categories and gpt-oss-120b in 2 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in math, where gpt-oss-120b leads 50.6 to 30.4.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 37.8% for DeepSeek-V3 and 88.9% for gpt-oss-120b.
- gpt-oss-120b is cheaper at $0.037 / $0.17 per million input/output tokens, against $0.24 / $0.90 for DeepSeek-V3.
- DeepSeek-V3 accepts more context: 164K tokens versus 131K.
Side by side
| DeepSeek-V3 | gpt-oss-120b | |
|---|---|---|
| Provider | DeepSeek | OpenAI |
| Noometry Index | 39.2 | 36.0 |
| Released | 2024-12-26 | 2025-08-05 |
| Weights | Open | Open |
| Context window | 164K | 131K |
| Max output | 164K | 41K |
| Input $ / M tokens | $0.24 | $0.037 |
| Output $ / M tokens | $0.90 | $0.17 |
| Results tracked | 60 | 48 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding DeepSeek-V3 leads
DeepSeek-V3: 41.9 (#99), gpt-oss-120b: 33.2 (#251)
| Benchmark | DeepSeek-V3 | gpt-oss-120b |
|---|---|---|
| Aider Polyglot | 55.1% | 41.8% |
| SciCode | 35.8% | 36% |
| WeirdML | 36.1% | 48.2% |
| LMArena Coding | 1368 | 1380 |
| SWE-bench Verified (bash only) | — | 26% |
| BigCodeBench Instruct | 50% | — |
| LiveBench Coding | 70.9% | — |
| BigCodeBench Complete | 62.2% | — |
| ALE-Bench | — | 575.62 |
| AlgoTune | — | 1.41 |
| HumanEval+ | 86.6% | — |
| MBPP+ | 73% | — |
Agentic & Tool Use Not comparable
DeepSeek-V3: —, gpt-oss-120b: 12.1 (#145)
| Benchmark | DeepSeek-V3 | gpt-oss-120b |
|---|---|---|
| METR Time Horizons | 49.6% | 56.6% |
| Terminal-Bench | — | 18.7% |
| APEX-Agents | — | 4.4% |
| Vending-Bench 2 | — | -21.53 |
Reasoning Too close to call
DeepSeek-V3: 20.0 (#226), gpt-oss-120b: 19.5 (#236)
| Benchmark | DeepSeek-V3 | gpt-oss-120b |
|---|---|---|
| SimpleBench | 27.2% | 22.1% |
| Kagi LLM Benchmark | 52.3% | 58.6% |
| CritPt | 0% | 1.1% |
| LMArena Hard Prompts | 1365 | 1364 |
| DTBench | 64.8% | 76.3% |
| LMCA | 15.5% | 22.1% |
| Epoch Capabilities Index | 135.95 | 139.94 |
| Chess Puzzles | — | 20% |
| LiveBench Reasoning | 65.8% | — |
| Mystery Game Puzzles | — | 2% |
| LiveBench Data Analysis | 60.9% | — |
| Surface Evolver Bench | — | 25% |
| BIG-Bench Hard | 87.5% | — |
| ForecastBench | 59.1 | — |
| HellaSwag | 88.9% | — |
| LiveBench | 66.9% | — |
| PIQA | 84.7% | — |
| WinoGrande | 85.2% | — |
Math gpt-oss-120b leads
DeepSeek-V3: 30.4 (#208), gpt-oss-120b: 50.6 (#43)
| Benchmark | DeepSeek-V3 | gpt-oss-120b |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 37.8% | 88.9% |
| Omni-MATH | 40.3% | 68.8% |
| LMArena Math | 1373 | 1388 |
| LiveBench Math | 73.5% | — |
| MATH Level 5 | 75.5% | — |
| FrontierMath (Feb 2025 set) | 1.7% | — |
Knowledge gpt-oss-120b leads
DeepSeek-V3: 37.5 (#147), gpt-oss-120b: 42.3 (#89)
| Benchmark | DeepSeek-V3 | gpt-oss-120b |
|---|---|---|
| GPQA Diamond | 67.6% | 75.8% |
| MMLU-Pro | 72.3% | 79.5% |
| Confabulations | 26.1% | 15.7% |
| Vectara Hallucination Rate | 6.1% | 14.2% |
| GPQA (HELM) | 53.8% | 68.4% |
| LMArena Expert | 1352 | 1356 |
| ARC (AI2) Challenge | 95.3% | — |
| MMLU | 87.2% | — |
| TriviaQA | 82.9% | — |
Multilingual Too close to call
DeepSeek-V3: 48.8 (#138), gpt-oss-120b: 48.2 (#142)
| Benchmark | DeepSeek-V3 | gpt-oss-120b |
|---|---|---|
| LMArena Non-English | 1358 | 1350 |
| LMArena Chinese | 1390 | 1384 |
| LMArena French | 1386 | 1370 |
| LMArena German | 1374 | 1355 |
| LMArena Japanese | 1333 | 1332 |
| LMArena Korean | 1321 | 1283 |
| LMArena Russian | 1372 | 1343 |
| LMArena Spanish | 1358 | 1388 |
Instruction Following DeepSeek-V3 leads
DeepSeek-V3: 73.0 (#125), gpt-oss-120b: 69.5 (#168)
| Benchmark | DeepSeek-V3 | gpt-oss-120b |
|---|---|---|
| IFEval | 83.2% | 83.6% |
| LMArena Instruction Following | 1345 | 1318 |
| LiveBench Instruction Following | 81.5% | — |
Long Context DeepSeek-V3 leads
DeepSeek-V3: 34.0 (#248), gpt-oss-120b: 31.4 (#273)
| Benchmark | DeepSeek-V3 | gpt-oss-120b |
|---|---|---|
| Fiction.LiveBench | 50% | 44.4% |
| LMArena Longer Query | 1353 | 1319 |
Writing & Preference DeepSeek-V3 leads
DeepSeek-V3: 57.6 (#124), gpt-oss-120b: 46.7 (#213)
| Benchmark | DeepSeek-V3 | gpt-oss-120b |
|---|---|---|
| LMArena Text | 1375 | 1366 |
| LMArena Creative Writing | 1364 | 1274 |
| Short-Story Creative Writing | 77% | 77.1% |
| EQ-Bench Creative Writing | 1472 | 961 |
| WildBench | 83% | 84.5% |
| LMArena Multi-Turn | 1389 | 1340 |
| LiveBench Language | 49.1% | — |
Frequently asked questions
Is DeepSeek-V3 better than gpt-oss-120b?
DeepSeek-V3 is the stronger model overall, scoring 39.2 to 36.0 on the Noometry Index. gpt-oss-120b costs 5.8× less per token, which makes it the better buy when DeepSeek-V3's lead doesn't matter for your workload.
Which is cheaper, DeepSeek-V3 or gpt-oss-120b?
gpt-oss-120b is cheaper. It lists at $0.037 per million input tokens and $0.17 per million output tokens; DeepSeek-V3 lists at $0.24 and $0.90.
Is DeepSeek-V3 or gpt-oss-120b better for coding?
DeepSeek-V3 scores higher on coding benchmarks: 41.9 versus 33.2 in the Noometry coding category.
Which has the bigger context window?
DeepSeek-V3 does, with 164K tokens against 131K.
How many benchmarks do DeepSeek-V3 and gpt-oss-120b share?
39 benchmarks have published results for both models. DeepSeek-V3 has 60 scored results on Noometry and gpt-oss-120b has 48.