Model comparison
Claude Sonnet 4.5 vs o3
o3 is the stronger model overall, scoring 47.5 to 44.1 on the Noometry Index.
Last verified . 55 shared benchmarks.
Summary
- They share 55 benchmarks with published results for both. Claude Sonnet 4.5 scores higher in 5 categories and o3 in 5 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where o3 leads 50.2 to 32.3.
- The biggest single-benchmark swing is OSWorld: 62.9% for Claude Sonnet 4.5 and 23% for o3.
- o3 is cheaper at $2 / $8 per million input/output tokens, against $3 / $15 for Claude Sonnet 4.5.
Side by side
| Claude Sonnet 4.5 | o3 | |
|---|---|---|
| Provider | Anthropic | OpenAI |
| Noometry Index | 44.1 | 47.5 |
| Released | 2025-09-29 | 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 | 73 | 63 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Too close to call
Claude Sonnet 4.5: 47.3 (#61), o3: 46.8 (#64)
| Benchmark | Claude Sonnet 4.5 | o3 |
|---|---|---|
| SWE-bench Verified | 71.3% | 62.3% |
| SWE-bench Verified (bash only) | 71.4% | 58.4% |
| GSO | 14.7% | 8.8% |
| WeirdML | 47.7% | 52.4% |
| LMArena Coding | 1489 | 1408 |
| ALE-Bench | 796.15 | 933.55 |
| Aider Polyglot | — | 81.3% |
| LMArena WebDev | 1393 | — |
| SWE-bench Multilingual | 67% | — |
| SciCode | 44.7% | — |
| CadEval | — | 74% |
| AlgoTune | 1.52 | — |
Agentic & Tool Use Claude Sonnet 4.5 leads
Claude Sonnet 4.5: 38.3 (#32), o3: 34.5 (#44)
| Benchmark | Claude Sonnet 4.5 | o3 |
|---|---|---|
| Berkeley Function Calling Leaderboard | 73.2% | 63% |
| GDPval | 42.5% | 30.8% |
| DeepResearch Bench | 52.6% | 45.2% |
| OSWorld | 62.9% | 23% |
| LMArena Search | 1159 | 1144 |
| METR Time Horizons | 67.4% | 65.4% |
| Terminal-Bench | 46.5% | — |
| Remote Labor Index | 2.1% | — |
| τ²-bench Airline | 72% | — |
| τ²-bench Banking | 25.3% | — |
| τ²-bench Retail | 72.4% | — |
| τ²-bench Telecom | 84.9% | — |
| Cybench | 60% | — |
| Vending-Bench 2 | 3,839 | — |
Reasoning o3 leads
Claude Sonnet 4.5: 26.9 (#125), o3: 32.0 (#78)
| Benchmark | Claude Sonnet 4.5 | o3 |
|---|---|---|
| ARC-AGI-2 | 13.6% | 6.5% |
| SimpleBench | 54.3% | 53.1% |
| Kagi LLM Benchmark | 57.9% | 67.6% |
| ARC-AGI-1 | 63.7% | 60.8% |
| CritPt | 1.1% | 1.4% |
| Chess Puzzles | 12% | 38% |
| EnigmaEval | 6% | 13.1% |
| LMArena Hard Prompts | 1462 | 1402 |
| Mystery Game Puzzles | 17% | 29% |
| DTBench | 83.2% | 84.8% |
| LMCA | 38.8% | 39.7% |
| Epoch Capabilities Index | 146.84 | 146.86 |
| ForecastBench | 61.9 | 62.5 |
| NYT Connections (extended) | 37.3% | — |
| EBR-Bench | 2.4% | — |
Math o3 leads
Claude Sonnet 4.5: 32.3 (#216), o3: 50.2 (#58)
| Benchmark | Claude Sonnet 4.5 | o3 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 23.9% | 33.3% |
| OTIS Mock AIME 2024-2025 | 77.8% | 84.4% |
| Omni-MATH | 55.3% | 71.4% |
| LMArena Math | 1449 | 1426 |
| MATH Level 5 | 97.7% | 97.8% |
| FrontierMath (Feb 2025 set) | 15.2% | 18.7% |
| FrontierMath Tier 4 (v1) | 4.2% | 2.1% |
| FrontierMath Tier 4 | 2.4% | — |
| ProofBench | 19% | — |
Knowledge o3 leads
Claude Sonnet 4.5: 48.4 (#76), o3: 54.6 (#52)
| Benchmark | Claude Sonnet 4.5 | o3 |
|---|---|---|
| GPQA Diamond | 82.3% | 81.8% |
| Humanity's Last Exam | 13.7% | 20.3% |
| SimpleQA Verified | 30.7% | 49.4% |
| MMLU-Pro | 86.9% | 85.9% |
| GPQA (HELM) | 68.6% | 75.3% |
| LMArena Expert | 1482 | 1402 |
| Confabulations | — | 14.4% |
| Vectara Hallucination Rate | 12% | — |
Multimodal o3 leads
Claude Sonnet 4.5: 34.8 (#89), o3: 41.4 (#36)
| Benchmark | Claude Sonnet 4.5 | o3 |
|---|---|---|
| VPCT | 39.8% | 52% |
| LMArena Vision | — | 1214 |
| GeoBench | — | 74% |
| LMArena Document | 1450 | — |
Multilingual Claude Sonnet 4.5 leads
Claude Sonnet 4.5: 53.4 (#69), o3: 51.7 (#105)
| Benchmark | Claude Sonnet 4.5 | o3 |
|---|---|---|
| LMArena Non-English | 1425 | 1401 |
| LMArena Chinese | 1459 | 1437 |
| LMArena French | 1458 | 1430 |
| LMArena German | 1427 | 1420 |
| LMArena Japanese | 1390 | 1403 |
| LMArena Korean | 1403 | 1370 |
| LMArena Russian | 1437 | 1406 |
| LMArena Spanish | 1457 | 1395 |
Instruction Following Claude Sonnet 4.5 leads
Claude Sonnet 4.5: 75.0 (#78), o3: 72.8 (#127)
| Benchmark | Claude Sonnet 4.5 | o3 |
|---|---|---|
| IFEval | 85% | 86.9% |
| LMArena Instruction Following | 1459 | 1368 |
Long Context o3 leads
Claude Sonnet 4.5: 45.2 (#46), o3: 53.3 (#6)
| Benchmark | Claude Sonnet 4.5 | o3 |
|---|---|---|
| LMArena Longer Query | 1476 | 1372 |
| Fiction.LiveBench | — | 88.9% |
| CL-bench | — | 17.8% |
Writing & Preference Claude Sonnet 4.5 leads
Claude Sonnet 4.5: 66.5 (#34), o3: 63.5 (#64)
| Benchmark | Claude Sonnet 4.5 | o3 |
|---|---|---|
| LMArena Text | 1439 | 1410 |
| LMArena Creative Writing | 1442 | 1359 |
| EQ-Bench Creative Writing | 1678 | 1676 |
| WildBench | 85.4% | 86.1% |
| LMArena Multi-Turn | 1465 | 1405 |
| Short-Story Creative Writing | — | 83.9% |
Frequently asked questions
Is Claude Sonnet 4.5 better than o3?
o3 is the stronger model overall, scoring 47.5 to 44.1 on the Noometry Index.
Which is cheaper, Claude Sonnet 4.5 or o3?
o3 is cheaper. It lists at $2 per million input tokens and $8 per million output tokens; Claude Sonnet 4.5 lists at $3 and $15.
Is Claude Sonnet 4.5 or o3 better for coding?
They score almost the same on coding (47.3 vs 46.8); test both on your own repository before choosing.
Which has the bigger context window?
Both accept 200K tokens.
How many benchmarks do Claude Sonnet 4.5 and o3 share?
55 benchmarks have published results for both models. Claude Sonnet 4.5 has 73 scored results on Noometry and o3 has 63.