Model comparison
Claude Sonnet 4 vs GPT-5 Nano
Claude Sonnet 4 is the stronger model overall, scoring 40.6 to 33.4 on the Noometry Index. GPT-5 Nano costs 44× less per token, which makes it the better buy when Claude Sonnet 4's lead doesn't matter for your workload.
Last verified . 40 shared benchmarks.
Summary
- They share 40 benchmarks with published results for both. Claude Sonnet 4 scores higher in 8 categories and GPT-5 Nano in 2 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in writing & preference, where Claude Sonnet 4 leads 57.3 to 39.4.
- The biggest single-benchmark swing is SWE-bench Verified (bash only): 64.9% for Claude Sonnet 4 and 34.8% for GPT-5 Nano.
- GPT-5 Nano is cheaper at $0.05 / $0.40 per million input/output tokens, against $3 / $15 for Claude Sonnet 4.
- GPT-5 Nano accepts more context: 400K tokens versus 200K.
Side by side
| Claude Sonnet 4 | GPT-5 Nano | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 40.6 | 33.4 |
| Released | 2025-05-22 | 2025-08-07 |
| Weights | Proprietary | Proprietary |
| Context window | 200K | 400K |
| Max output | 64K | 128K |
| Input $ / M tokens | $3 | $0.05 |
| Output $ / M tokens | $15 | $0.40 |
| Results tracked | 58 | 48 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Claude Sonnet 4 leads
Claude Sonnet 4: 44.0 (#70), GPT-5 Nano: 33.3 (#248)
| Benchmark | Claude Sonnet 4 | GPT-5 Nano |
|---|---|---|
| SWE-bench Verified (bash only) | 64.9% | 34.8% |
| WeirdML | 46.1% | 38.1% |
| LMArena Coding | 1414 | 1351 |
| ALE-Bench | 655.35 | 718.67 |
| Aider Polyglot | 61.3% | — |
| SciCode | 40% | — |
| GSO | 4.9% | — |
Agentic & Tool Use Claude Sonnet 4 leads
Claude Sonnet 4: 38.3 (#29), GPT-5 Nano: 25.6 (#99)
| Benchmark | Claude Sonnet 4 | GPT-5 Nano |
|---|---|---|
| Terminal-Bench | — | 21.8% |
| Berkeley Function Calling Leaderboard | — | 51.5% |
| TheAgentCompany | 33.1% | — |
| Cybench | 35% | — |
| DeepResearch Bench | 46.6% | — |
| OSWorld | 43.9% | — |
| METR Time Horizons | 62% | — |
Reasoning Claude Sonnet 4 leads
Claude Sonnet 4: 23.0 (#169), GPT-5 Nano: 16.4 (#291)
| Benchmark | Claude Sonnet 4 | GPT-5 Nano |
|---|---|---|
| ARC-AGI-2 | 5.9% | 2.6% |
| Kagi LLM Benchmark | 73% | 62.2% |
| ARC-AGI-1 | 40% | 20.7% |
| LMArena Hard Prompts | 1372 | 1329 |
| DTBench | 77.1% | 62.7% |
| LMCA | 29% | 7.9% |
| Epoch Capabilities Index | 141.69 | 139.39 |
| ForecastBench | 60.2 | 59.1 |
| SimpleBench | 45.5% | — |
| CritPt | 0.3% | — |
| Chess Puzzles | — | 27% |
| EnigmaEval | 3.1% | — |
| Mystery Game Puzzles | — | 9% |
Math Claude Sonnet 4 leads
Claude Sonnet 4: 41.5 (#71), GPT-5 Nano: 29.1 (#221)
| Benchmark | Claude Sonnet 4 | GPT-5 Nano |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 71.1% | 81.1% |
| Omni-MATH | 60.2% | 54.6% |
| LMArena Math | 1375 | 1318 |
| MATH Level 5 | 84.4% | 95.2% |
| FrontierMath (Feb 2025 set) | 4.1% | 8.3% |
| FrontierMath Tier 4 (v1) | 0% | 2.1% |
| FrontierMath (Tiers 1-3) | — | 20% |
| FrontierMath Tier 4 | — | 2.4% |
| ProofBench | — | 12% |
Knowledge Claude Sonnet 4 leads
Claude Sonnet 4: 41.6 (#99), GPT-5 Nano: 35.8 (#166)
| Benchmark | Claude Sonnet 4 | GPT-5 Nano |
|---|---|---|
| GPQA Diamond | 79.2% | 69.4% |
| MMLU-Pro | 84.3% | 77.8% |
| Vectara Hallucination Rate | 10.3% | 10.5% |
| GPQA (HELM) | 70.6% | 67.9% |
| LMArena Expert | 1371 | 1321 |
| Humanity's Last Exam | 7.8% | — |
| SimpleQA Verified | — | 11.7% |
| Confabulations | 13.2% | — |
Multimodal GPT-5 Nano leads
Claude Sonnet 4: 25.6 (#116), GPT-5 Nano: 30.7 (#102)
| Benchmark | Claude Sonnet 4 | GPT-5 Nano |
|---|---|---|
| LMArena Vision | 1191 | 1159 |
| VPCT | 34% | 37.2% |
| GeoBench | 37% | — |
| MindCube | 44.8% | — |
Multilingual Claude Sonnet 4 leads
Claude Sonnet 4: 46.9 (#151), GPT-5 Nano: 45.5 (#167)
| Benchmark | Claude Sonnet 4 | GPT-5 Nano |
|---|---|---|
| LMArena Non-English | 1333 | 1313 |
| LMArena Chinese | 1351 | 1355 |
| LMArena German | 1331 | 1329 |
| LMArena Korean | 1293 | 1269 |
| LMArena Russian | 1355 | 1296 |
| LMArena Spanish | 1356 | 1360 |
| LMArena French | 1364 | — |
| LMArena Japanese | 1301 | — |
Instruction Following GPT-5 Nano leads
Claude Sonnet 4: 71.8 (#141), GPT-5 Nano: 75.2 (#72)
| Benchmark | Claude Sonnet 4 | GPT-5 Nano |
|---|---|---|
| IFEval | 84% | 93.2% |
| LMArena Instruction Following | 1375 | 1306 |
Long Context Claude Sonnet 4 leads
Claude Sonnet 4: 33.6 (#254), GPT-5 Nano: 31.3 (#275)
| Benchmark | Claude Sonnet 4 | GPT-5 Nano |
|---|---|---|
| Fiction.LiveBench | 46.9% | 44.4% |
| LMArena Longer Query | 1398 | 1313 |
Writing & Preference Claude Sonnet 4 leads
Claude Sonnet 4: 57.3 (#127), GPT-5 Nano: 39.4 (#244)
| Benchmark | Claude Sonnet 4 | GPT-5 Nano |
|---|---|---|
| LMArena Text | 1351 | 1320 |
| LMArena Creative Writing | 1345 | 1250 |
| EQ-Bench Creative Writing | 1483 | 705 |
| WildBench | 83.8% | 80.6% |
| LMArena Multi-Turn | 1376 | 1311 |
| Short-Story Creative Writing | 81.4% | — |
Frequently asked questions
Is Claude Sonnet 4 better than GPT-5 Nano?
Claude Sonnet 4 is the stronger model overall, scoring 40.6 to 33.4 on the Noometry Index. GPT-5 Nano costs 44× less per token, which makes it the better buy when Claude Sonnet 4's lead doesn't matter for your workload.
Which is cheaper, Claude Sonnet 4 or GPT-5 Nano?
GPT-5 Nano is cheaper. It lists at $0.05 per million input tokens and $0.40 per million output tokens; Claude Sonnet 4 lists at $3 and $15.
Is Claude Sonnet 4 or GPT-5 Nano better for coding?
Claude Sonnet 4 scores higher on coding benchmarks: 44.0 versus 33.3 in the Noometry coding category.
Which has the bigger context window?
GPT-5 Nano does, with 400K tokens against 200K.
How many benchmarks do Claude Sonnet 4 and GPT-5 Nano share?
40 benchmarks have published results for both models. Claude Sonnet 4 has 58 scored results on Noometry and GPT-5 Nano has 48.