Model comparison
GPT-5.4 mini vs gpt-oss-120b
GPT-5.4 mini is the stronger model overall, scoring 44.8 to 36.0 on the Noometry Index. gpt-oss-120b costs 24× less per token, which makes it the better buy when GPT-5.4 mini's lead doesn't matter for your workload.
Last verified . 31 shared benchmarks.
Summary
- They share 31 benchmarks with published results for both. GPT-5.4 mini scores higher in 8 categories and gpt-oss-120b in 1 category; 9 gaps are clear of the uncertainty.
- The widest gap is in agentic & tool use, where GPT-5.4 mini leads 29.6 to 12.1.
- The biggest single-benchmark swing is Kagi LLM Benchmark: 37.9% for GPT-5.4 mini and 58.6% for gpt-oss-120b.
- gpt-oss-120b is cheaper at $0.037 / $0.17 per million input/output tokens, against $0.75 / $4.50 for GPT-5.4 mini.
- GPT-5.4 mini accepts more context: 400K tokens versus 131K.
- gpt-oss-120b has downloadable open weights; the other is API-only.
Side by side
| GPT-5.4 mini | gpt-oss-120b | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 44.8 | 36.0 |
| Released | 2026-03-17 | 2025-08-05 |
| Weights | Proprietary | Open |
| Context window | 400K | 131K |
| Max output | 128K | 41K |
| Input $ / M tokens | $0.75 | $0.037 |
| Output $ / M tokens | $4.50 | $0.17 |
| Results tracked | 46 | 48 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.4 mini leads
GPT-5.4 mini: 45.0 (#67), gpt-oss-120b: 33.2 (#251)
| Benchmark | GPT-5.4 mini | gpt-oss-120b |
|---|---|---|
| SciCode | 49.9% | 36% |
| WeirdML | 60.3% | 48.2% |
| LMArena Coding | 1438 | 1380 |
| ALE-Bench | 1,189 | 575.62 |
| FrontierCode | 27% | — |
| SWE-bench Verified (bash only) | — | 26% |
| Aider Polyglot | — | 41.8% |
| LMArena WebDev | 1397 | — |
| AlgoTune | — | 1.41 |
Agentic & Tool Use GPT-5.4 mini leads
GPT-5.4 mini: 29.6 (#74), gpt-oss-120b: 12.1 (#145)
| Benchmark | GPT-5.4 mini | gpt-oss-120b |
|---|---|---|
| Terminal-Bench | — | 18.7% |
| APEX-Agents | — | 4.4% |
| DeepResearch Bench | 36.3% | — |
| METR Time Horizons | — | 56.6% |
| Vending-Bench 2 | — | -21.53 |
Reasoning GPT-5.4 mini leads
GPT-5.4 mini: 30.3 (#77), gpt-oss-120b: 19.5 (#236)
| Benchmark | GPT-5.4 mini | gpt-oss-120b |
|---|---|---|
| Kagi LLM Benchmark | 37.9% | 58.6% |
| CritPt | 10% | 1.1% |
| Chess Puzzles | 24% | 20% |
| LMArena Hard Prompts | 1424 | 1364 |
| Mystery Game Puzzles | 11% | 2% |
| DTBench | 80% | 76.3% |
| LMCA | 40.8% | 22.1% |
| Epoch Capabilities Index | 148.83 | 139.94 |
| ARC-AGI-2 | 18.9% | — |
| SimpleBench | — | 22.1% |
| NYT Connections (extended) | 61.8% | — |
| ARC-AGI-1 | 63.7% | — |
| Thematic Generalization | 61.7% | — |
| Surface Evolver Bench | — | 25% |
| ForecastBench | 57 | — |
Math gpt-oss-120b leads
GPT-5.4 mini: 45.4 (#63), gpt-oss-120b: 50.6 (#43)
| Benchmark | GPT-5.4 mini | gpt-oss-120b |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 88.9% | 88.9% |
| LMArena Math | 1418 | 1388 |
| FrontierMath (Tiers 1-3) | 51.2% | — |
| FrontierMath Tier 4 | 9.8% | — |
| ProofBench | 21% | — |
| Omni-MATH | — | 68.8% |
| FrontierMath (Feb 2025 set) | 28.3% | — |
| FrontierMath Tier 4 (v1) | 2.1% | — |
Knowledge GPT-5.4 mini leads
GPT-5.4 mini: 51.4 (#62), gpt-oss-120b: 42.3 (#89)
| Benchmark | GPT-5.4 mini | gpt-oss-120b |
|---|---|---|
| GPQA Diamond | 86.9% | 75.8% |
| Vectara Hallucination Rate | 5.5% | 14.2% |
| LMArena Expert | 1434 | 1356 |
| SimpleQA Verified | 29.4% | — |
| MMLU-Pro | — | 79.5% |
| Confabulations | — | 15.7% |
| GPQA (HELM) | — | 68.4% |
Multimodal Not comparable
GPT-5.4 mini: 39.0 (#51), gpt-oss-120b: —
| Benchmark | GPT-5.4 mini | gpt-oss-120b |
|---|---|---|
| LMArena Vision | 1245 | — |
Multilingual GPT-5.4 mini leads
GPT-5.4 mini: 52.1 (#94), gpt-oss-120b: 48.2 (#142)
| Benchmark | GPT-5.4 mini | gpt-oss-120b |
|---|---|---|
| LMArena Non-English | 1405 | 1350 |
| LMArena Chinese | 1446 | 1384 |
| LMArena French | 1440 | 1370 |
| LMArena German | 1407 | 1355 |
| LMArena Japanese | 1375 | 1332 |
| LMArena Korean | 1369 | 1283 |
| LMArena Russian | 1417 | 1343 |
| LMArena Spanish | 1403 | 1388 |
Instruction Following GPT-5.4 mini leads
GPT-5.4 mini: 74.3 (#99), gpt-oss-120b: 69.5 (#168)
| Benchmark | GPT-5.4 mini | gpt-oss-120b |
|---|---|---|
| LMArena Instruction Following | 1405 | 1318 |
| IFEval | — | 83.6% |
Long Context GPT-5.4 mini leads
GPT-5.4 mini: 42.9 (#107), gpt-oss-120b: 31.4 (#273)
| Benchmark | GPT-5.4 mini | gpt-oss-120b |
|---|---|---|
| LMArena Longer Query | 1407 | 1319 |
| Fiction.LiveBench | — | 44.4% |
Writing & Preference GPT-5.4 mini leads
GPT-5.4 mini: 64.2 (#56), gpt-oss-120b: 46.7 (#213)
| Benchmark | GPT-5.4 mini | gpt-oss-120b |
|---|---|---|
| LMArena Text | 1411 | 1366 |
| LMArena Creative Writing | 1370 | 1274 |
| EQ-Bench Creative Writing | 1665 | 961 |
| LMArena Multi-Turn | 1429 | 1340 |
| Short-Story Creative Writing | — | 77.1% |
| WildBench | — | 84.5% |
Frequently asked questions
Is GPT-5.4 mini better than gpt-oss-120b?
GPT-5.4 mini is the stronger model overall, scoring 44.8 to 36.0 on the Noometry Index. gpt-oss-120b costs 24× less per token, which makes it the better buy when GPT-5.4 mini's lead doesn't matter for your workload.
Which is cheaper, GPT-5.4 mini 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-5.4 mini lists at $0.75 and $4.50.
Is GPT-5.4 mini or gpt-oss-120b better for coding?
GPT-5.4 mini scores higher on coding benchmarks: 45.0 versus 33.2 in the Noometry coding category.
Which has the bigger context window?
GPT-5.4 mini does, with 400K tokens against 131K.
How many benchmarks do GPT-5.4 mini and gpt-oss-120b share?
31 benchmarks have published results for both models. GPT-5.4 mini has 46 scored results on Noometry and gpt-oss-120b has 48.