Model comparison
GPT-5 vs gpt-oss-120b
GPT-5 is the stronger model overall, scoring 50.9 to 36.3 on the Noometry Index. gpt-oss-120b costs 49× less per token, which makes it the better buy when GPT-5's lead doesn't matter for your workload.
Last verified . 45 shared benchmarks.
Summary
- They share 45 benchmarks with published results for both. GPT-5 scores higher in 9 categories and gpt-oss-120b in 0 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in long context, where GPT-5 leads 69.5 to 31.4.
- The biggest single-benchmark swing is Fiction.LiveBench: 97.2% for GPT-5 and 44.4% for gpt-oss-120b.
- gpt-oss-120b is cheaper at $0.037 / $0.17 per million input/output tokens, against $1.25 / $10 for GPT-5.
- GPT-5 accepts more context: 400K tokens versus 131K.
- gpt-oss-120b has downloadable open weights; the other is API-only.
Side by side
| GPT-5 | gpt-oss-120b | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 50.9 | 36.3 |
| Released | 2025-08-07 | 2025-08-05 |
| Weights | Proprietary | Open |
| Context window | 400K | 131K |
| Max output | 128K | 41K |
| Input $ / M tokens | $1.25 | $0.037 |
| Output $ / M tokens | $10 | $0.17 |
| Results tracked | 69 | 48 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5 leads
GPT-5: 50.3 (#47), gpt-oss-120b: 33.5 (#256)
| Benchmark | GPT-5 | gpt-oss-120b |
|---|---|---|
| SWE-bench Verified (bash only) | 65% | 26% |
| Aider Polyglot | 88% | 41.8% |
| SciCode | 42.9% | 36% |
| WeirdML | 60.7% | 48.2% |
| LMArena Coding | 1436 | 1380 |
| ALE-Bench | 1,162 | 575.62 |
| AlgoTune | 1.67 | 1.41 |
| SWE-bench Verified | 73.6% | — |
| LMArena WebDev | 1418 | — |
| GSO | 6.9% | — |
Agentic & Tool Use GPT-5 leads
GPT-5: 33.1 (#56), gpt-oss-120b: 12.2 (#153)
| Benchmark | GPT-5 | gpt-oss-120b |
|---|---|---|
| Terminal-Bench | 49.6% | 18.7% |
| METR Time Horizons | 69.6% | 56.6% |
| APEX-Agents | — | 4.4% |
| GDPval | 34.8% | — |
| Remote Labor Index | 1.7% | — |
| DeepResearch Bench | 49.6% | — |
| BALROG | 32.8% | — |
| LMArena Search | 1133 | — |
| Vending-Bench 2 | — | -21.53 |
Reasoning GPT-5 leads
GPT-5: 38.3 (#64), gpt-oss-120b: 20.0 (#245)
| Benchmark | GPT-5 | gpt-oss-120b |
|---|---|---|
| SimpleBench | 56.7% | 22.1% |
| Kagi LLM Benchmark | 72.7% | 58.6% |
| CritPt | 12.6% | 1.1% |
| Chess Puzzles | 37% | 20% |
| LMArena Hard Prompts | 1416 | 1364 |
| Mystery Game Puzzles | 23% | 2% |
| DTBench | 90.7% | 76.3% |
| LMCA | 40% | 22.1% |
| Epoch Capabilities Index | 150 | 139.93 |
| ARC-AGI-2 | 9.9% | — |
| ARC-AGI-1 | 65.7% | — |
| EnigmaEval | 10.5% | — |
| EBR-Bench | 12.7% | — |
| Surface Evolver Bench | — | 25% |
| ForecastBench | 61.4 | — |
Math GPT-5 leads
GPT-5: 55.0 (#44), gpt-oss-120b: 52.5 (#50)
| Benchmark | GPT-5 | gpt-oss-120b |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 91.4% | 88.9% |
| Omni-MATH | 64.7% | 68.8% |
| LMArena Math | 1407 | 1389 |
| FrontierMath (Tiers 1-3) | 55.4% | — |
| FrontierMath Tier 4 | 22% | — |
| ProofBench | 18% | — |
| MATH Level 5 | 98.1% | — |
| FrontierMath (Feb 2025 set) | 32.4% | — |
| FrontierMath Tier 4 (v1) | 12.5% | — |
Knowledge GPT-5 leads
GPT-5: 56.6 (#43), gpt-oss-120b: 42.4 (#96)
| Benchmark | GPT-5 | gpt-oss-120b |
|---|---|---|
| GPQA Diamond | 86.2% | 75.8% |
| MMLU-Pro | 86.3% | 79.5% |
| Confabulations | 10.3% | 15.7% |
| Vectara Hallucination Rate | 14.7% | 14.2% |
| GPQA (HELM) | 79.2% | 68.4% |
| LMArena Expert | 1419 | 1356 |
| Humanity's Last Exam | 25.3% | — |
| SimpleQA Verified | 50.1% | — |
Multimodal Not comparable
GPT-5: 46.8 (#13), gpt-oss-120b: —
| Benchmark | GPT-5 | gpt-oss-120b |
|---|---|---|
| LMArena Vision | 1232 | — |
| GeoBench | 81% | — |
| VPCT | 66% | — |
Multilingual GPT-5 leads
GPT-5: 51.4 (#110), gpt-oss-120b: 48.0 (#147)
| Benchmark | GPT-5 | gpt-oss-120b |
|---|---|---|
| LMArena Non-English | 1397 | 1351 |
| LMArena Chinese | 1422 | 1385 |
| LMArena French | 1410 | 1369 |
| LMArena German | 1416 | 1353 |
| LMArena Japanese | 1409 | 1331 |
| LMArena Korean | 1360 | 1282 |
| LMArena Russian | 1406 | 1343 |
| LMArena Spanish | 1399 | 1389 |
Instruction Following GPT-5 leads
GPT-5: 73.8 (#113), gpt-oss-120b: 69.3 (#173)
| Benchmark | GPT-5 | gpt-oss-120b |
|---|---|---|
| IFEval | 87.5% | 83.6% |
| LMArena Instruction Following | 1388 | 1318 |
Long Context GPT-5 leads
GPT-5: 69.5 (#2), gpt-oss-120b: 31.4 (#278)
| Benchmark | GPT-5 | gpt-oss-120b |
|---|---|---|
| Fiction.LiveBench | 97.2% | 44.4% |
| LMArena Longer Query | 1399 | 1319 |
Writing & Preference GPT-5 leads
GPT-5: 63.4 (#65), gpt-oss-120b: 46.5 (#217)
| Benchmark | GPT-5 | gpt-oss-120b |
|---|---|---|
| LMArena Text | 1406 | 1365 |
| LMArena Creative Writing | 1365 | 1275 |
| Short-Story Creative Writing | 86% | 77.1% |
| EQ-Bench Creative Writing | 1627 | 961 |
| WildBench | 85.7% | 84.5% |
| LMArena Multi-Turn | 1426 | 1340 |
Frequently asked questions
Is GPT-5 better than gpt-oss-120b?
GPT-5 is the stronger model overall, scoring 50.9 to 36.3 on the Noometry Index. gpt-oss-120b costs 49× less per token, which makes it the better buy when GPT-5's lead doesn't matter for your workload.
Which is cheaper, GPT-5 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; GPT-5 lists at $1.25 and $10.
Is GPT-5 or gpt-oss-120b better for coding?
GPT-5 scores higher on coding benchmarks: 50.3 versus 33.5 in the Noometry coding category.
Which has the bigger context window?
GPT-5 does, with 400K tokens against 131K.
How many benchmarks do GPT-5 and gpt-oss-120b share?
45 benchmarks have published results for both models. GPT-5 has 69 scored results on Noometry and gpt-oss-120b has 48.