Model comparison
gpt-oss-120b vs Llama 3.1-70B
gpt-oss-120b is the stronger model overall, scoring 36.0 to 29.4 on the Noometry Index.
Last verified . 29 shared benchmarks.
MetaSummary
- They share 29 benchmarks with published results for both. gpt-oss-120b scores higher in 6 categories and Llama 3.1-70B in 3 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where gpt-oss-120b leads 50.6 to 12.7.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 88.9% for gpt-oss-120b and 3.6% for Llama 3.1-70B.
- gpt-oss-120b is cheaper at $0.037 / $0.17 per million input/output tokens, against $0.40 / $0.40 for Llama 3.1-70B.
- gpt-oss-120b accepts more context: 131K tokens versus 128K.
Side by side
| gpt-oss-120b | Llama 3.1-70B | |
|---|---|---|
| Provider | OpenAI | Meta |
| Noometry Index | 36.0 | 29.4 |
| Released | 2025-08-05 | 2024-07-23 |
| Weights | Open | Open |
| Context window | 131K | 128K |
| Max output | 41K | 4K |
| Input $ / M tokens | $0.037 | $0.40 |
| Output $ / M tokens | $0.17 | $0.40 |
| Results tracked | 48 | 35 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding gpt-oss-120b leads
gpt-oss-120b: 33.2 (#251), Llama 3.1-70B: 29.8 (#292)
| Benchmark | gpt-oss-120b | Llama 3.1-70B |
|---|---|---|
| WeirdML | 48.2% | 9% |
| LMArena Coding | 1380 | 1260 |
| SWE-bench Verified (bash only) | 26% | — |
| Aider Polyglot | 41.8% | — |
| SciCode | 36% | — |
| BigCodeBench Instruct | — | 46.1% |
| BigCodeBench Complete | — | 54.8% |
| ALE-Bench | 575.62 | — |
| AlgoTune | 1.41 | — |
Agentic & Tool Use Llama 3.1-70B leads
gpt-oss-120b: 12.1 (#145), Llama 3.1-70B: 24.9 (#104)
| Benchmark | gpt-oss-120b | Llama 3.1-70B |
|---|---|---|
| Terminal-Bench | 18.7% | — |
| APEX-Agents | 4.4% | — |
| TheAgentCompany | — | 6.9% |
| BALROG | — | 27.9% |
| METR Time Horizons | 56.6% | — |
| Vending-Bench 2 | -21.53 | — |
Reasoning Llama 3.1-70B leads
gpt-oss-120b: 19.5 (#236), Llama 3.1-70B: 20.8 (#215)
| Benchmark | gpt-oss-120b | Llama 3.1-70B |
|---|---|---|
| LMArena Hard Prompts | 1364 | 1241 |
| DTBench | 76.3% | 60% |
| LMCA | 22.1% | 14.8% |
| Epoch Capabilities Index | 139.94 | 125.91 |
| SimpleBench | 22.1% | — |
| Kagi LLM Benchmark | 58.6% | — |
| CritPt | 1.1% | — |
| Chess Puzzles | 20% | — |
| Mystery Game Puzzles | 2% | — |
| Surface Evolver Bench | 25% | — |
Math gpt-oss-120b leads
gpt-oss-120b: 50.6 (#43), Llama 3.1-70B: 12.7 (#298)
| Benchmark | gpt-oss-120b | Llama 3.1-70B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 88.9% | 3.6% |
| Omni-MATH | 68.8% | 21% |
| LMArena Math | 1388 | 1252 |
| MATH Level 5 | — | 36.7% |
Knowledge gpt-oss-120b leads
gpt-oss-120b: 42.3 (#89), Llama 3.1-70B: 24.1 (#264)
| Benchmark | gpt-oss-120b | Llama 3.1-70B |
|---|---|---|
| GPQA Diamond | 75.8% | 44.2% |
| MMLU-Pro | 79.5% | 65.3% |
| GPQA (HELM) | 68.4% | 42.6% |
| LMArena Expert | 1356 | 1209 |
| Confabulations | 15.7% | — |
| Vectara Hallucination Rate | 14.2% | — |
| MMLU | — | 80.1% |
Multilingual gpt-oss-120b leads
gpt-oss-120b: 48.2 (#142), Llama 3.1-70B: 38.9 (#220)
| Benchmark | gpt-oss-120b | Llama 3.1-70B |
|---|---|---|
| LMArena Non-English | 1350 | 1219 |
| LMArena Chinese | 1384 | 1215 |
| LMArena French | 1370 | 1260 |
| LMArena German | 1355 | 1222 |
| LMArena Japanese | 1332 | 1132 |
| LMArena Korean | 1283 | 1140 |
| LMArena Russian | 1343 | 1234 |
| LMArena Spanish | 1388 | 1253 |
Instruction Following gpt-oss-120b leads
gpt-oss-120b: 69.5 (#168), Llama 3.1-70B: 65.5 (#218)
| Benchmark | gpt-oss-120b | Llama 3.1-70B |
|---|---|---|
| IFEval | 83.6% | 82.1% |
| LMArena Instruction Following | 1318 | 1231 |
Long Context Llama 3.1-70B leads
gpt-oss-120b: 31.4 (#273), Llama 3.1-70B: 37.6 (#208)
| Benchmark | gpt-oss-120b | Llama 3.1-70B |
|---|---|---|
| LMArena Longer Query | 1319 | 1241 |
| Fiction.LiveBench | 44.4% | — |
Writing & Preference gpt-oss-120b leads
gpt-oss-120b: 46.7 (#213), Llama 3.1-70B: 35.6 (#262)
| Benchmark | gpt-oss-120b | Llama 3.1-70B |
|---|---|---|
| LMArena Text | 1366 | 1261 |
| LMArena Creative Writing | 1274 | 1232 |
| EQ-Bench Creative Writing | 961 | 784 |
| WildBench | 84.5% | 75.8% |
| LMArena Multi-Turn | 1340 | 1256 |
| Short-Story Creative Writing | 77.1% | — |
Frequently asked questions
Is gpt-oss-120b better than Llama 3.1-70B?
gpt-oss-120b is the stronger model overall, scoring 36.0 to 29.4 on the Noometry Index.
Which is cheaper, gpt-oss-120b or Llama 3.1-70B?
gpt-oss-120b is cheaper. It lists at $0.037 per million input tokens and $0.17 per million output tokens; Llama 3.1-70B lists at $0.40 and $0.40.
Is gpt-oss-120b or Llama 3.1-70B better for coding?
gpt-oss-120b scores higher on coding benchmarks: 33.2 versus 29.8 in the Noometry coding category.
Which has the bigger context window?
gpt-oss-120b does, with 131K tokens against 128K.
How many benchmarks do gpt-oss-120b and Llama 3.1-70B share?
29 benchmarks have published results for both models. gpt-oss-120b has 48 scored results on Noometry and Llama 3.1-70B has 35.