Model comparison
Claude Sonnet 4 vs GPT-5.4 nano
GPT-5.4 nano is the stronger model overall, scoring 41.9 to 40.6 on the Noometry Index.
Last verified . 34 shared benchmarks.
Summary
- They share 34 benchmarks with published results for both. Claude Sonnet 4 scores higher in 3 categories and GPT-5.4 nano in 6 categories; 4 gaps are clear of the uncertainty.
- The widest gap is in multimodal, where GPT-5.4 nano leads 36.0 to 25.6.
- The biggest single-benchmark swing is Kagi LLM Benchmark: 73% for Claude Sonnet 4 and 39.7% for GPT-5.4 nano.
- GPT-5.4 nano is cheaper at $0.20 / $1.25 per million input/output tokens, against $3 / $15 for Claude Sonnet 4.
- GPT-5.4 nano accepts more context: 400K tokens versus 200K.
Side by side
| Claude Sonnet 4 | GPT-5.4 nano | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 40.6 | 41.9 |
| Released | 2025-05-22 | 2026-03-17 |
| Weights | Proprietary | Proprietary |
| Context window | 200K | 400K |
| Max output | 64K | 128K |
| Input $ / M tokens | $3 | $0.20 |
| Output $ / M tokens | $15 | $1.25 |
| Results tracked | 58 | 40 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Too close to call
Claude Sonnet 4: 44.0 (#70), GPT-5.4 nano: 43.1 (#82)
| Benchmark | Claude Sonnet 4 | GPT-5.4 nano |
|---|---|---|
| SciCode | 40% | 46.9% |
| WeirdML | 46.1% | 49.2% |
| LMArena Coding | 1414 | 1404 |
| ALE-Bench | 655.35 | 1,005 |
| SWE-bench Verified (bash only) | 64.9% | — |
| Aider Polyglot | 61.3% | — |
| GSO | 4.9% | — |
Agentic & Tool Use Not comparable
Claude Sonnet 4: 38.3 (#29), GPT-5.4 nano: —
| Benchmark | Claude Sonnet 4 | GPT-5.4 nano |
|---|---|---|
| TheAgentCompany | 33.1% | — |
| Cybench | 35% | — |
| DeepResearch Bench | 46.6% | — |
| OSWorld | 43.9% | — |
| METR Time Horizons | 62% | — |
Reasoning Too close to call
Claude Sonnet 4: 23.0 (#169), GPT-5.4 nano: 23.8 (#159)
| Benchmark | Claude Sonnet 4 | GPT-5.4 nano |
|---|---|---|
| ARC-AGI-2 | 5.9% | 5.7% |
| Kagi LLM Benchmark | 73% | 39.7% |
| ARC-AGI-1 | 40% | 51.5% |
| CritPt | 0.3% | 9.3% |
| LMArena Hard Prompts | 1372 | 1380 |
| DTBench | 77.1% | 80.3% |
| LMCA | 29% | 36.9% |
| Epoch Capabilities Index | 141.69 | 145.78 |
| ForecastBench | 60.2 | 57.3 |
| SimpleBench | 45.5% | — |
| Chess Puzzles | — | 30% |
| EnigmaEval | 3.1% | — |
| Mystery Game Puzzles | — | 9% |
Math Too close to call
Claude Sonnet 4: 41.5 (#71), GPT-5.4 nano: 40.7 (#76)
| Benchmark | Claude Sonnet 4 | GPT-5.4 nano |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 71.1% | 87.8% |
| LMArena Math | 1375 | 1406 |
| FrontierMath (Feb 2025 set) | 4.1% | 25.9% |
| FrontierMath Tier 4 (v1) | 0% | 6.3% |
| FrontierMath (Tiers 1-3) | — | 44.9% |
| FrontierMath Tier 4 | — | 12.2% |
| ProofBench | — | 5% |
| Omni-MATH | 60.2% | — |
| MATH Level 5 | 84.4% | — |
Knowledge Too close to call
Claude Sonnet 4: 41.6 (#99), GPT-5.4 nano: 41.8 (#94)
| Benchmark | Claude Sonnet 4 | GPT-5.4 nano |
|---|---|---|
| GPQA Diamond | 79.2% | 78.5% |
| Vectara Hallucination Rate | 10.3% | 3.1% |
| LMArena Expert | 1371 | 1397 |
| Humanity's Last Exam | 7.8% | — |
| SimpleQA Verified | — | 11.7% |
| MMLU-Pro | 84.3% | — |
| Confabulations | 13.2% | — |
| GPQA (HELM) | 70.6% | — |
Multimodal GPT-5.4 nano leads
Claude Sonnet 4: 25.6 (#116), GPT-5.4 nano: 36.0 (#73)
| Benchmark | Claude Sonnet 4 | GPT-5.4 nano |
|---|---|---|
| LMArena Vision | 1191 | 1196 |
| GeoBench | 37% | — |
| VPCT | 34% | — |
| MindCube | 44.8% | — |
Multilingual GPT-5.4 nano leads
Claude Sonnet 4: 46.9 (#151), GPT-5.4 nano: 48.8 (#136)
| Benchmark | Claude Sonnet 4 | GPT-5.4 nano |
|---|---|---|
| LMArena Non-English | 1333 | 1359 |
| LMArena Chinese | 1351 | 1391 |
| LMArena French | 1364 | 1396 |
| LMArena German | 1331 | 1368 |
| LMArena Japanese | 1301 | 1342 |
| LMArena Korean | 1293 | 1321 |
| LMArena Russian | 1355 | 1363 |
| LMArena Spanish | 1356 | 1368 |
Instruction Following Too close to call
Claude Sonnet 4: 71.8 (#141), GPT-5.4 nano: 72.1 (#139)
| Benchmark | Claude Sonnet 4 | GPT-5.4 nano |
|---|---|---|
| LMArena Instruction Following | 1375 | 1362 |
| IFEval | 84% | — |
Long Context GPT-5.4 nano leads
Claude Sonnet 4: 33.6 (#254), GPT-5.4 nano: 41.6 (#131)
| Benchmark | Claude Sonnet 4 | GPT-5.4 nano |
|---|---|---|
| LMArena Longer Query | 1398 | 1365 |
| Fiction.LiveBench | 46.9% | — |
Writing & Preference Claude Sonnet 4 leads
Claude Sonnet 4: 57.3 (#127), GPT-5.4 nano: 55.9 (#137)
| Benchmark | Claude Sonnet 4 | GPT-5.4 nano |
|---|---|---|
| LMArena Text | 1351 | 1372 |
| LMArena Creative Writing | 1345 | 1313 |
| LMArena Multi-Turn | 1376 | 1382 |
| Short-Story Creative Writing | 81.4% | — |
| EQ-Bench Creative Writing | 1483 | — |
| WildBench | 83.8% | — |
Frequently asked questions
Is Claude Sonnet 4 better than GPT-5.4 nano?
GPT-5.4 nano is the stronger model overall, scoring 41.9 to 40.6 on the Noometry Index.
Which is cheaper, Claude Sonnet 4 or GPT-5.4 nano?
GPT-5.4 nano is cheaper. It lists at $0.20 per million input tokens and $1.25 per million output tokens; Claude Sonnet 4 lists at $3 and $15.
Is Claude Sonnet 4 or GPT-5.4 nano better for coding?
They score almost the same on coding (44.0 vs 43.1); test both on your own repository before choosing.
Which has the bigger context window?
GPT-5.4 nano does, with 400K tokens against 200K.
How many benchmarks do Claude Sonnet 4 and GPT-5.4 nano share?
34 benchmarks have published results for both models. Claude Sonnet 4 has 58 scored results on Noometry and GPT-5.4 nano has 40.