Model comparison
GPT-5.4 mini vs gpt-oss-20b
GPT-5.4 mini is the stronger model overall, scoring 45.0 to 32.5 on the Noometry Index. gpt-oss-20b costs 47× less per token, which makes it the better buy when GPT-5.4 mini's lead doesn't matter for your workload.
Last verified . 28 shared benchmarks.
Summary
- They share 28 benchmarks with published results for both. GPT-5.4 mini scores higher in 9 categories and gpt-oss-20b in 0 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where GPT-5.4 mini leads 64.0 to 35.5.
- The biggest single-benchmark swing is LMCA: 40.8% for GPT-5.4 mini and 14.5% for gpt-oss-20b.
- gpt-oss-20b is cheaper at $0.018 / $0.09 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-20b has downloadable open weights; the other is API-only.
Side by side
| GPT-5.4 mini | gpt-oss-20b | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 45.0 | 32.5 |
| Released | 2026-03-17 | 2025-08-05 |
| Weights | Proprietary | Open |
| Context window | 400K | 131K |
| Max output | 128K | 16K |
| Input $ / M tokens | $0.75 | $0.018 |
| Output $ / M tokens | $4.50 | $0.09 |
| Results tracked | 46 | 34 |
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: 45.2 (#72), gpt-oss-20b: 37.6 (#192)
| Benchmark | GPT-5.4 mini | gpt-oss-20b |
|---|---|---|
| SciCode | 49.9% | 34.4% |
| WeirdML | 60.3% | 40.9% |
| LMArena Coding | 1438 | 1306 |
| ALE-Bench | 1,189 | 566.05 |
| FrontierCode | 27% | — |
| LMArena WebDev | 1397 | — |
Agentic & Tool Use GPT-5.4 mini leads
GPT-5.4 mini: 29.9 (#81), gpt-oss-20b: 9.3 (#154)
| Benchmark | GPT-5.4 mini | gpt-oss-20b |
|---|---|---|
| Terminal-Bench | — | 3.4% |
| DeepResearch Bench | 36.3% | — |
Reasoning GPT-5.4 mini leads
GPT-5.4 mini: 30.4 (#85), gpt-oss-20b: 19.3 (#261)
| Benchmark | GPT-5.4 mini | gpt-oss-20b |
|---|---|---|
| Kagi LLM Benchmark | 37.9% | 53.2% |
| CritPt | 10% | 1.4% |
| Chess Puzzles | 24% | 4% |
| LMArena Hard Prompts | 1424 | 1274 |
| DTBench | 80% | 68% |
| LMCA | 40.8% | 14.5% |
| Epoch Capabilities Index | 148.84 | 137.82 |
| ARC-AGI-2 | 18.9% | — |
| NYT Connections (extended) | 61.8% | — |
| ARC-AGI-1 | 63.7% | — |
| Thematic Generalization | 61.7% | — |
| Mystery Game Puzzles | 11% | — |
| ForecastBench | 57 | — |
Math GPT-5.4 mini leads
GPT-5.4 mini: 45.5 (#75), gpt-oss-20b: 39.4 (#103)
| Benchmark | GPT-5.4 mini | gpt-oss-20b |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 88.9% | 65.3% |
| LMArena Math | 1419 | 1317 |
| FrontierMath (Tiers 1-3) | 51.2% | — |
| FrontierMath Tier 4 | 9.8% | — |
| ProofBench | 21% | — |
| Omni-MATH | — | 56.5% |
| FrontierMath (Feb 2025 set) | 28.3% | — |
| FrontierMath Tier 4 (v1) | 2.1% | — |
Knowledge GPT-5.4 mini leads
GPT-5.4 mini: 51.5 (#67), gpt-oss-20b: 34.6 (#195)
| Benchmark | GPT-5.4 mini | gpt-oss-20b |
|---|---|---|
| GPQA Diamond | 86.9% | 60.8% |
| LMArena Expert | 1435 | 1258 |
| SimpleQA Verified | 29.4% | — |
| MMLU-Pro | — | 74% |
| Vectara Hallucination Rate | 5.5% | — |
| GPQA (HELM) | — | 59.4% |
Multimodal Not comparable
GPT-5.4 mini: 39.7 (#56), gpt-oss-20b: —
| Benchmark | GPT-5.4 mini | gpt-oss-20b |
|---|---|---|
| LMArena Vision | 1245 | — |
Multilingual GPT-5.4 mini leads
GPT-5.4 mini: 51.9 (#96), gpt-oss-20b: 42.2 (#197)
| Benchmark | GPT-5.4 mini | gpt-oss-20b |
|---|---|---|
| LMArena Non-English | 1405 | 1268 |
| LMArena Chinese | 1446 | 1314 |
| LMArena German | 1409 | 1255 |
| LMArena Japanese | 1374 | 1244 |
| LMArena Korean | 1368 | 1236 |
| LMArena Russian | 1417 | 1278 |
| LMArena Spanish | 1405 | 1267 |
| LMArena French | 1440 | — |
Instruction Following GPT-5.4 mini leads
GPT-5.4 mini: 74.1 (#102), gpt-oss-20b: 61.8 (#240)
| Benchmark | GPT-5.4 mini | gpt-oss-20b |
|---|---|---|
| LMArena Instruction Following | 1405 | 1236 |
| IFEval | — | 73.2% |
Long Context GPT-5.4 mini leads
GPT-5.4 mini: 43.0 (#112), gpt-oss-20b: 37.9 (#209)
| Benchmark | GPT-5.4 mini | gpt-oss-20b |
|---|---|---|
| LMArena Longer Query | 1407 | 1250 |
Writing & Preference GPT-5.4 mini leads
GPT-5.4 mini: 64.0 (#58), gpt-oss-20b: 35.5 (#265)
| Benchmark | GPT-5.4 mini | gpt-oss-20b |
|---|---|---|
| LMArena Text | 1412 | 1287 |
| LMArena Creative Writing | 1370 | 1201 |
| EQ-Bench Creative Writing | 1665 | 666 |
| LMArena Multi-Turn | 1429 | 1268 |
| WildBench | — | 73.7% |
Frequently asked questions
Is GPT-5.4 mini better than gpt-oss-20b?
GPT-5.4 mini is the stronger model overall, scoring 45.0 to 32.5 on the Noometry Index. gpt-oss-20b costs 47× 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-20b?
gpt-oss-20b is cheaper. It lists at $0.018 per million input tokens and $0.09 per million output tokens; GPT-5.4 mini lists at $0.75 and $4.50.
Is GPT-5.4 mini or gpt-oss-20b better for coding?
GPT-5.4 mini scores higher on coding benchmarks: 45.2 versus 37.6 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-20b share?
28 benchmarks have published results for both models. GPT-5.4 mini has 46 scored results on Noometry and gpt-oss-20b has 34.