Model comparison
Claude Sonnet 4 vs o3
o3 is the stronger model overall, scoring 47.5 to 40.8 on the Noometry Index.
Last verified . 53 shared benchmarks.
Summary
- They share 53 benchmarks with published results for both. Claude Sonnet 4 scores higher in 1 category and o3 in 9 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in long context, where o3 leads 53.3 to 33.7.
- The biggest single-benchmark swing is Fiction.LiveBench: 46.9% for Claude Sonnet 4 and 88.9% for o3.
- o3 is cheaper at $2 / $8 per million input/output tokens, against $3 / $15 for Claude Sonnet 4.
Side by side
| Claude Sonnet 4 | o3 | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 40.8 | 47.5 |
| Released | 2025-05-22 | 2025-04-16 |
| Weights | Proprietary | Proprietary |
| Context window | 200K | 200K |
| Max output | 64K | 100K |
| Input $ / M tokens | $3 | $2 |
| Output $ / M tokens | $15 | $8 |
| Results tracked | 58 | 63 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding o3 leads
Claude Sonnet 4: 43.5 (#88), o3: 46.8 (#64)
| Benchmark | Claude Sonnet 4 | o3 |
|---|---|---|
| SWE-bench Verified (bash only) | 64.9% | 58.4% |
| Aider Polyglot | 61.3% | 81.3% |
| GSO | 4.9% | 8.8% |
| WeirdML | 46.1% | 52.4% |
| LMArena Coding | 1414 | 1408 |
| ALE-Bench | 655.35 | 933.55 |
| SWE-bench Verified | — | 62.3% |
| SciCode | 40% | — |
| CadEval | — | 74% |
Agentic & Tool Use Claude Sonnet 4 leads
Claude Sonnet 4: 38.5 (#31), o3: 34.5 (#44)
| Benchmark | Claude Sonnet 4 | o3 |
|---|---|---|
| DeepResearch Bench | 46.6% | 45.2% |
| OSWorld | 43.9% | 23% |
| METR Time Horizons | 62% | 65.4% |
| Berkeley Function Calling Leaderboard | — | 63% |
| GDPval | — | 30.8% |
| TheAgentCompany | 33.1% | — |
| Cybench | 35% | — |
| LMArena Search | — | 1144 |
Reasoning o3 leads
Claude Sonnet 4: 22.9 (#187), o3: 32.0 (#78)
| Benchmark | Claude Sonnet 4 | o3 |
|---|---|---|
| ARC-AGI-2 | 5.9% | 6.5% |
| SimpleBench | 45.5% | 53.1% |
| Kagi LLM Benchmark | 73% | 67.6% |
| ARC-AGI-1 | 40% | 60.8% |
| CritPt | 0.3% | 1.4% |
| EnigmaEval | 3.1% | 13.1% |
| LMArena Hard Prompts | 1372 | 1402 |
| DTBench | 77.1% | 84.8% |
| LMCA | 29% | 39.7% |
| Epoch Capabilities Index | 141.69 | 146.86 |
| ForecastBench | 60.2 | 62.5 |
| Chess Puzzles | — | 38% |
| Mystery Game Puzzles | — | 29% |
Math o3 leads
Claude Sonnet 4: 43.3 (#80), o3: 50.2 (#58)
| Benchmark | Claude Sonnet 4 | o3 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 71.1% | 84.4% |
| Omni-MATH | 60.2% | 71.4% |
| LMArena Math | 1375 | 1426 |
| MATH Level 5 | 84.4% | 97.8% |
| FrontierMath (Feb 2025 set) | 4.1% | 18.7% |
| FrontierMath Tier 4 (v1) | 0% | 2.1% |
| FrontierMath (Tiers 1-3) | — | 33.3% |
Knowledge o3 leads
Claude Sonnet 4: 41.8 (#108), o3: 54.6 (#52)
| Benchmark | Claude Sonnet 4 | o3 |
|---|---|---|
| GPQA Diamond | 79.2% | 81.8% |
| Humanity's Last Exam | 7.8% | 20.3% |
| MMLU-Pro | 84.3% | 85.9% |
| Confabulations | 13.2% | 14.4% |
| GPQA (HELM) | 70.6% | 75.3% |
| LMArena Expert | 1372 | 1402 |
| SimpleQA Verified | — | 49.4% |
| Vectara Hallucination Rate | 10.3% | — |
Multimodal o3 leads
Claude Sonnet 4: 26.2 (#121), o3: 41.4 (#36)
| Benchmark | Claude Sonnet 4 | o3 |
|---|---|---|
| LMArena Vision | 1191 | 1214 |
| GeoBench | 37% | 74% |
| VPCT | 34% | 52% |
| MindCube | 44.8% | — |
Multilingual o3 leads
Claude Sonnet 4: 46.7 (#156), o3: 51.7 (#105)
| Benchmark | Claude Sonnet 4 | o3 |
|---|---|---|
| LMArena Non-English | 1333 | 1401 |
| LMArena Chinese | 1350 | 1437 |
| LMArena French | 1363 | 1430 |
| LMArena German | 1331 | 1420 |
| LMArena Japanese | 1302 | 1403 |
| LMArena Korean | 1291 | 1370 |
| LMArena Russian | 1355 | 1406 |
| LMArena Spanish | 1357 | 1395 |
Instruction Following o3 leads
Claude Sonnet 4: 71.7 (#145), o3: 72.8 (#127)
| Benchmark | Claude Sonnet 4 | o3 |
|---|---|---|
| IFEval | 84% | 86.9% |
| LMArena Instruction Following | 1376 | 1368 |
Long Context o3 leads
Claude Sonnet 4: 33.7 (#259), o3: 53.3 (#6)
| Benchmark | Claude Sonnet 4 | o3 |
|---|---|---|
| Fiction.LiveBench | 46.9% | 88.9% |
| LMArena Longer Query | 1398 | 1372 |
| CL-bench | — | 17.8% |
Writing & Preference o3 leads
Claude Sonnet 4: 57.1 (#132), o3: 63.5 (#64)
| Benchmark | Claude Sonnet 4 | o3 |
|---|---|---|
| LMArena Text | 1351 | 1410 |
| LMArena Creative Writing | 1345 | 1359 |
| Short-Story Creative Writing | 81.4% | 83.9% |
| EQ-Bench Creative Writing | 1483 | 1676 |
| WildBench | 83.8% | 86.1% |
| LMArena Multi-Turn | 1376 | 1405 |
Frequently asked questions
Is Claude Sonnet 4 better than o3?
o3 is the stronger model overall, scoring 47.5 to 40.8 on the Noometry Index.
Which is cheaper, Claude Sonnet 4 or o3?
o3 is cheaper. It lists at $2 per million input tokens and $8 per million output tokens; Claude Sonnet 4 lists at $3 and $15.
Is Claude Sonnet 4 or o3 better for coding?
o3 scores higher on coding benchmarks: 46.8 versus 43.5 in the Noometry coding category.
Which has the bigger context window?
Both accept 200K tokens.
How many benchmarks do Claude Sonnet 4 and o3 share?
53 benchmarks have published results for both models. Claude Sonnet 4 has 58 scored results on Noometry and o3 has 63.