Model comparison
GPT-5.1 vs GPT-5.4
GPT-5.4 is the stronger model overall, scoring 59.4 to 49.0 on the Noometry Index. GPT-5.1 costs 1.6× less per token, which makes it the better buy when GPT-5.4's lead doesn't matter for your workload.
Last verified . 48 shared benchmarks.
Summary
- They share 48 benchmarks with published results for both. GPT-5.1 scores higher in 2 categories and GPT-5.4 in 8 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where GPT-5.4 leads 61.8 to 39.8.
- The biggest single-benchmark swing is ARC-AGI-2: 17.6% for GPT-5.1 and 74% for GPT-5.4.
- GPT-5.1 is cheaper at $1.25 / $10 per million input/output tokens, against $2.50 / $15 for GPT-5.4.
- GPT-5.4 accepts more context: 1.05M tokens versus 400K.
Side by side
| GPT-5.1 | GPT-5.4 | |
|---|---|---|
| Provider | OpenAI | OpenAI |
| Noometry Index | 49.0 | 59.4 |
| Released | 2025-11-13 | 2026-03-05 |
| Weights | Proprietary | Proprietary |
| Context window | 400K | 1.05M |
| Max output | 128K | 128K |
| Input $ / M tokens | $1.25 | $2.50 |
| Output $ / M tokens | $10 | $15 |
| Results tracked | 63 | 68 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding GPT-5.4 leads
GPT-5.1: 46.4 (#66), GPT-5.4: 52.6 (#33)
| Benchmark | GPT-5.1 | GPT-5.4 |
|---|---|---|
| SWE-bench Verified | 68% | 76.9% |
| LMArena WebDev | 1395 | 1465 |
| SciCode | 43.3% | 56.6% |
| GSO | 13.7% | 31.4% |
| WeirdML | 60.8% | 77.7% |
| LMArena Coding | 1454 | 1497 |
| ALE-Bench | 1,192 | 1,607 |
| DeepSWE | — | 51.8% |
| SWE-bench Verified (bash only) | 66% | — |
| LiveBench Coding | 72.5% | — |
| MirrorCode | — | 15.6% |
| AlgoTune | — | 1.85 |
Agentic & Tool Use GPT-5.4 leads
GPT-5.1: 32.7 (#60), GPT-5.4: 46.5 (#13)
| Benchmark | GPT-5.1 | GPT-5.4 |
|---|---|---|
| Terminal-Bench | 47.6% | 81.8% |
| DeepResearch Bench | 42.8% | 35.1% |
| LMArena Search | 1199 | 1197 |
| Vending-Bench 2 | 1,473 | 6,144 |
| APEX-Agents | — | 52.4% |
| τ²-bench Banking | — | 39.4% |
| PostTrainBench | — | 19% |
| GBAEval | — | 45.1% |
| METR Time Horizons | — | 74.3% |
Reasoning GPT-5.4 leads
GPT-5.1: 39.8 (#58), GPT-5.4: 61.8 (#19)
| Benchmark | GPT-5.1 | GPT-5.4 |
|---|---|---|
| ARC-AGI-2 | 17.6% | 74% |
| ARC-AGI-1 | 72.8% | 93.7% |
| CritPt | 4.9% | 23.4% |
| Chess Puzzles | 32% | 44% |
| EnigmaEval | 11.2% | 16% |
| LMArena Hard Prompts | 1457 | 1485 |
| Mystery Game Puzzles | 19% | 37% |
| DTBench | 90.1% | 94.4% |
| LMCA | 43.9% | 52% |
| Epoch Capabilities Index | 149.64 | 156.81 |
| ForecastBench | 58.1 | 59.5 |
| SimpleBench | 53.2% | — |
| Kagi LLM Benchmark | — | 63.8% |
| NYT Connections (extended) | — | 91.3% |
| Thematic Generalization | — | 80% |
| EBR-Bench | — | 25.4% |
| LiveBench Reasoning | 95.8% | — |
| LiveBench Data Analysis | 72.1% | — |
| LiveBench | 78.8% | — |
Math GPT-5.4 leads
GPT-5.1: 52.2 (#51), GPT-5.4: 73.5 (#19)
| Benchmark | GPT-5.1 | GPT-5.4 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 88.6% | 97.8% |
| LMArena Math | 1447 | 1488 |
| FrontierMath (Feb 2025 set) | 31% | 47.6% |
| FrontierMath Tier 4 (v1) | 12.5% | 27.1% |
| FrontierMath (Tiers 1-3) | — | 78.6% |
| FrontierMath Tier 4 | — | 49% |
| MathArena Final-Answer Competitions | — | 83.1% |
| ProofBench | — | 56% |
| Omni-MATH | 46.4% | — |
| LiveBench Math | 94.5% | — |
Knowledge GPT-5.4 leads
GPT-5.1: 50.6 (#71), GPT-5.4: 65.3 (#14)
| Benchmark | GPT-5.1 | GPT-5.4 |
|---|---|---|
| GPQA Diamond | 87.6% | 93.3% |
| Humanity's Last Exam | 23.7% | 36.2% |
| SimpleQA Verified | 48% | 45.1% |
| Vectara Hallucination Rate | 10.9% | 7% |
| LMArena Expert | 1470 | 1507 |
| MMLU-Pro | 57.9% | — |
| GPQA (HELM) | 44.2% | — |
Multimodal GPT-5.1 leads
GPT-5.1: 44.8 (#19), GPT-5.4: 43.7 (#20)
| Benchmark | GPT-5.1 | GPT-5.4 |
|---|---|---|
| LMArena Vision | 1250 | 1303 |
| LMArena Document | 1403 | 1471 |
| VPCT | 58.7% | — |
| Blueprint-Bench 2 | — | 27.1% |
| Furniture Assembly | — | 37.5% |
Multilingual GPT-5.4 leads
GPT-5.1: 53.8 (#56), GPT-5.4: 56.2 (#23)
| Benchmark | GPT-5.1 | GPT-5.4 |
|---|---|---|
| LMArena Non-English | 1431 | 1465 |
| LMArena Chinese | 1495 | 1519 |
| LMArena French | 1450 | 1493 |
| LMArena German | 1438 | 1472 |
| LMArena Japanese | 1453 | 1485 |
| LMArena Korean | 1401 | 1448 |
| LMArena Russian | 1435 | 1480 |
| LMArena Spanish | 1433 | 1454 |
Instruction Following GPT-5.1 leads
GPT-5.1: 83.9 (#1), GPT-5.4: 77.1 (#27)
| Benchmark | GPT-5.1 | GPT-5.4 |
|---|---|---|
| LMArena Instruction Following | 1443 | 1469 |
| LiveBench Instruction Following | 93.3% | — |
| IFEval | 93.5% | — |
Long Context GPT-5.4 leads
GPT-5.1: 47.6 (#14), GPT-5.4: 50.3 (#8)
| Benchmark | GPT-5.1 | GPT-5.4 |
|---|---|---|
| CL-bench | 23.7% | 27.9% |
| CL-bench Life | 17.3% | 21.7% |
| LMArena Longer Query | 1447 | 1473 |
Writing & Preference GPT-5.4 leads
GPT-5.1: 64.5 (#55), GPT-5.4: 71.9 (#17)
| Benchmark | GPT-5.1 | GPT-5.4 |
|---|---|---|
| LMArena Text | 1443 | 1469 |
| LMArena Creative Writing | 1427 | 1439 |
| LMArena Multi-Turn | 1450 | 1482 |
| EQ-Bench Creative Writing | — | 1840 |
| WildBench | 86.3% | — |
| EQ-Bench 4 | — | 1272 |
| LiveBench Language | 80.2% | — |
Frequently asked questions
Is GPT-5.1 better than GPT-5.4?
GPT-5.4 is the stronger model overall, scoring 59.4 to 49.0 on the Noometry Index. GPT-5.1 costs 1.6× less per token, which makes it the better buy when GPT-5.4's lead doesn't matter for your workload.
Which is cheaper, GPT-5.1 or GPT-5.4?
GPT-5.1 is cheaper. It lists at $1.25 per million input tokens and $10 per million output tokens; GPT-5.4 lists at $2.50 and $15.
Is GPT-5.1 or GPT-5.4 better for coding?
GPT-5.4 scores higher on coding benchmarks: 52.6 versus 46.4 in the Noometry coding category.
Which has the bigger context window?
GPT-5.4 does, with 1.05M tokens against 400K.
How many benchmarks do GPT-5.1 and GPT-5.4 share?
48 benchmarks have published results for both models. GPT-5.1 has 63 scored results on Noometry and GPT-5.4 has 68.