Model comparison
gpt-oss-120b vs Llama 3-8B
gpt-oss-120b is the stronger model overall, scoring 36.0 to 25.3 on the Noometry Index.
Last verified . 22 shared benchmarks.
MetaSummary
- They share 22 benchmarks with published results for both. gpt-oss-120b scores higher in 7 categories and Llama 3-8B in 1 category; 8 gaps are clear of the uncertainty.
- The widest gap is in math, where gpt-oss-120b leads 50.6 to 8.3.
- The biggest single-benchmark swing is OTIS Mock AIME 2024-2025: 88.9% for gpt-oss-120b and 1.9% for Llama 3-8B.
- gpt-oss-120b is cheaper at $0.037 / $0.17 per million input/output tokens, against $0.14 / $0.14 for Llama 3-8B.
- gpt-oss-120b accepts more context: 131K tokens versus 8K.
Side by side
| gpt-oss-120b | Llama 3-8B | |
|---|---|---|
| Provider | OpenAI | Meta |
| Noometry Index | 36.0 | 25.3 |
| Released | 2025-08-05 | 2024-04-18 |
| Weights | Open | Open |
| Context window | 131K | 8K |
| Max output | 41K | 8K |
| Input $ / M tokens | $0.037 | $0.14 |
| Output $ / M tokens | $0.17 | $0.14 |
| Results tracked | 48 | 34 |
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 3-8B: 30.5 (#285)
| Benchmark | gpt-oss-120b | Llama 3-8B |
|---|---|---|
| LMArena Coding | 1380 | 1152 |
| SWE-bench Verified (bash only) | 26% | — |
| Aider Polyglot | 41.8% | — |
| SciCode | 36% | — |
| WeirdML | 48.2% | — |
| BigCodeBench Instruct | — | 31.9% |
| BigCodeBench Complete | — | 36.9% |
| ALE-Bench | 575.62 | — |
| AlgoTune | 1.41 | — |
| HumanEval+ | — | 56.7% |
| MBPP+ | — | 54.8% |
Agentic & Tool Use Not comparable
gpt-oss-120b: 12.1 (#145), Llama 3-8B: —
| Benchmark | gpt-oss-120b | Llama 3-8B |
|---|---|---|
| Terminal-Bench | 18.7% | — |
| APEX-Agents | 4.4% | — |
| METR Time Horizons | 56.6% | — |
| Vending-Bench 2 | -21.53 | — |
Reasoning gpt-oss-120b leads
gpt-oss-120b: 19.5 (#236), Llama 3-8B: 13.8 (#321)
| Benchmark | gpt-oss-120b | Llama 3-8B |
|---|---|---|
| Chess Puzzles | 20% | 0% |
| LMArena Hard Prompts | 1364 | 1133 |
| DTBench | 76.3% | 43.9% |
| Epoch Capabilities Index | 139.94 | 116.33 |
| SimpleBench | 22.1% | — |
| Kagi LLM Benchmark | 58.6% | — |
| CritPt | 1.1% | — |
| Mystery Game Puzzles | 2% | — |
| LMCA | 22.1% | — |
| Surface Evolver Bench | 25% | — |
| Adversarial NLI | — | 57.3% |
| ForecastBench | — | 58.6 |
| WinoGrande | — | 75.7% |
Math gpt-oss-120b leads
gpt-oss-120b: 50.6 (#43), Llama 3-8B: 8.3 (#317)
| Benchmark | gpt-oss-120b | Llama 3-8B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 88.9% | 1.9% |
| LMArena Math | 1388 | 1151 |
| Omni-MATH | 68.8% | — |
| MATH Level 5 | — | 6.1% |
Knowledge gpt-oss-120b leads
gpt-oss-120b: 42.3 (#89), Llama 3-8B: 7.8 (#303)
| Benchmark | gpt-oss-120b | Llama 3-8B |
|---|---|---|
| GPQA Diamond | 75.8% | 26.1% |
| LMArena Expert | 1356 | 1113 |
| MMLU-Pro | 79.5% | — |
| Confabulations | 15.7% | — |
| Vectara Hallucination Rate | 14.2% | — |
| GPQA (HELM) | 68.4% | — |
| ARC (AI2) Challenge | — | 82.8% |
| MMLU | — | 68.8% |
| OpenBookQA | — | 82.6% |
| TriviaQA | — | 67.7% |
Multilingual gpt-oss-120b leads
gpt-oss-120b: 48.2 (#142), Llama 3-8B: 31.0 (#256)
| Benchmark | gpt-oss-120b | Llama 3-8B |
|---|---|---|
| LMArena Non-English | 1350 | 1098 |
| LMArena Chinese | 1384 | 1076 |
| LMArena French | 1370 | 1158 |
| LMArena German | 1355 | 1104 |
| LMArena Japanese | 1332 | 967 |
| LMArena Korean | 1283 | 1003 |
| LMArena Russian | 1343 | 1109 |
| LMArena Spanish | 1388 | 1172 |
Instruction Following gpt-oss-120b leads
gpt-oss-120b: 69.5 (#168), Llama 3-8B: 58.7 (#255)
| Benchmark | gpt-oss-120b | Llama 3-8B |
|---|---|---|
| LMArena Instruction Following | 1318 | 1127 |
| IFEval | 83.6% | — |
Long Context Llama 3-8B leads
gpt-oss-120b: 31.4 (#273), Llama 3-8B: 34.2 (#246)
| Benchmark | gpt-oss-120b | Llama 3-8B |
|---|---|---|
| LMArena Longer Query | 1319 | 1128 |
| Fiction.LiveBench | 44.4% | — |
Writing & Preference gpt-oss-120b leads
gpt-oss-120b: 46.7 (#213), Llama 3-8B: 37.7 (#251)
| Benchmark | gpt-oss-120b | Llama 3-8B |
|---|---|---|
| LMArena Text | 1366 | 1166 |
| LMArena Creative Writing | 1274 | 1150 |
| LMArena Multi-Turn | 1340 | 1152 |
| Short-Story Creative Writing | 77.1% | — |
| EQ-Bench Creative Writing | 961 | — |
| WildBench | 84.5% | — |
Frequently asked questions
Is gpt-oss-120b better than Llama 3-8B?
gpt-oss-120b is the stronger model overall, scoring 36.0 to 25.3 on the Noometry Index.
Which is cheaper, gpt-oss-120b or Llama 3-8B?
gpt-oss-120b is cheaper. It lists at $0.037 per million input tokens and $0.17 per million output tokens; Llama 3-8B lists at $0.14 and $0.14.
Is gpt-oss-120b or Llama 3-8B better for coding?
gpt-oss-120b scores higher on coding benchmarks: 33.2 versus 30.5 in the Noometry coding category.
Which has the bigger context window?
gpt-oss-120b does, with 131K tokens against 8K.
How many benchmarks do gpt-oss-120b and Llama 3-8B share?
22 benchmarks have published results for both models. gpt-oss-120b has 48 scored results on Noometry and Llama 3-8B has 34.