Model comparison
Claude Sonnet 4 vs Gemini 3 Pro
Gemini 3 Pro is the stronger model overall, scoring 54.7 to 40.6 on the Noometry Index.
Last verified . 46 shared benchmarks.
GoogleSummary
- They share 46 benchmarks with published results for both. Claude Sonnet 4 scores higher in 0 categories and Gemini 3 Pro in 10 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in multimodal, where Gemini 3 Pro leads 56.9 to 25.6.
- The biggest single-benchmark swing is VPCT: 34% for Claude Sonnet 4 and 91% for Gemini 3 Pro.
Side by side
| Claude Sonnet 4 | Gemini 3 Pro | |
|---|---|---|
| Provider | Anthropic | |
| Noometry Index | 40.6 | 54.7 |
| Released | 2025-05-22 | 2025-11-18 |
| Weights | Proprietary | Proprietary |
| Context window | 200K | — |
| Max output | 64K | — |
| Input $ / M tokens | $3 | — |
| Output $ / M tokens | $15 | — |
| Results tracked | 58 | 67 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Gemini 3 Pro leads
Claude Sonnet 4: 44.0 (#70), Gemini 3 Pro: 52.1 (#33)
| Benchmark | Claude Sonnet 4 | Gemini 3 Pro |
|---|---|---|
| SWE-bench Verified (bash only) | 64.9% | 74.2% |
| GSO | 4.9% | 18.6% |
| WeirdML | 46.1% | 69.9% |
| LMArena Coding | 1414 | 1482 |
| ALE-Bench | 655.35 | 1,177 |
| SWE-bench Verified | — | 72.9% |
| Aider Polyglot | 61.3% | — |
| LMArena WebDev | — | 1439 |
| SWE-bench Multilingual | — | 68.7% |
| SciCode | 40% | — |
| AlgoTune | — | 1.83 |
Agentic & Tool Use Gemini 3 Pro leads
Claude Sonnet 4: 38.3 (#29), Gemini 3 Pro: 40.5 (#21)
| Benchmark | Claude Sonnet 4 | Gemini 3 Pro |
|---|---|---|
| DeepResearch Bench | 46.6% | 46.3% |
| METR Time Horizons | 62% | 71% |
| Terminal-Bench | — | 69.4% |
| Berkeley Function Calling Leaderboard | — | 72.5% |
| GDPval | — | 40.3% |
| Remote Labor Index | — | 1.3% |
| TheAgentCompany | 33.1% | — |
| τ²-bench Airline | — | 80.5% |
| τ²-bench Banking | — | 18% |
| τ²-bench Retail | — | 75.9% |
| τ²-bench Telecom | — | 91% |
| Cybench | 35% | — |
| OSWorld | 43.9% | — |
| BALROG | — | 58.1% |
| LMArena Search | — | 1207 |
| Vending-Bench 2 | — | 5,478 |
Reasoning Gemini 3 Pro leads
Claude Sonnet 4: 23.0 (#169), Gemini 3 Pro: 52.5 (#28)
| Benchmark | Claude Sonnet 4 | Gemini 3 Pro |
|---|---|---|
| ARC-AGI-2 | 5.9% | 31.1% |
| SimpleBench | 45.5% | 76.4% |
| Kagi LLM Benchmark | 73% | 80.1% |
| ARC-AGI-1 | 40% | 75% |
| CritPt | 0.3% | 6.9% |
| EnigmaEval | 3.1% | 18.2% |
| LMArena Hard Prompts | 1372 | 1481 |
| Epoch Capabilities Index | 141.69 | 152.96 |
| ForecastBench | 60.2 | 61.2 |
| NYT Connections (extended) | — | 94.4% |
| Chess Puzzles | — | 31% |
| DTBench | 77.1% | — |
| LMCA | 29% | — |
Math Gemini 3 Pro leads
Claude Sonnet 4: 41.5 (#71), Gemini 3 Pro: 48.8 (#50)
| Benchmark | Claude Sonnet 4 | Gemini 3 Pro |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 71.1% | 91.4% |
| Omni-MATH | 60.2% | 55.5% |
| LMArena Math | 1375 | 1475 |
| FrontierMath (Feb 2025 set) | 4.1% | 37.6% |
| FrontierMath Tier 4 (v1) | 0% | 18.8% |
| MathArena Final-Answer Competitions | — | 67% |
| ProofBench | — | 20% |
| MATH Level 5 | 84.4% | — |
Knowledge Gemini 3 Pro leads
Claude Sonnet 4: 41.6 (#99), Gemini 3 Pro: 64.3 (#15)
| Benchmark | Claude Sonnet 4 | Gemini 3 Pro |
|---|---|---|
| GPQA Diamond | 79.2% | 92.6% |
| Humanity's Last Exam | 7.8% | 37.5% |
| MMLU-Pro | 84.3% | 90.3% |
| Vectara Hallucination Rate | 10.3% | 13.6% |
| GPQA (HELM) | 70.6% | 80.3% |
| LMArena Expert | 1371 | 1475 |
| Confabulations | 13.2% | — |
Multimodal Gemini 3 Pro leads
Claude Sonnet 4: 25.6 (#116), Gemini 3 Pro: 56.9 (#1)
| Benchmark | Claude Sonnet 4 | Gemini 3 Pro |
|---|---|---|
| LMArena Vision | 1191 | 1305 |
| GeoBench | 37% | 84% |
| VPCT | 34% | 91% |
| LMArena Document | — | 1434 |
| MindCube | 44.8% | — |
Multilingual Gemini 3 Pro leads
Claude Sonnet 4: 46.9 (#151), Gemini 3 Pro: 57.1 (#14)
| Benchmark | Claude Sonnet 4 | Gemini 3 Pro |
|---|---|---|
| LMArena Non-English | 1333 | 1474 |
| LMArena Chinese | 1351 | 1522 |
| LMArena French | 1364 | 1492 |
| LMArena German | 1331 | 1517 |
| LMArena Japanese | 1301 | 1513 |
| LMArena Korean | 1293 | 1449 |
| LMArena Russian | 1355 | 1493 |
| LMArena Spanish | 1356 | 1468 |
Instruction Following Gemini 3 Pro leads
Claude Sonnet 4: 71.8 (#141), Gemini 3 Pro: 76.4 (#42)
| Benchmark | Claude Sonnet 4 | Gemini 3 Pro |
|---|---|---|
| IFEval | 84% | 87.7% |
| LMArena Instruction Following | 1375 | 1458 |
Long Context Gemini 3 Pro leads
Claude Sonnet 4: 33.6 (#254), Gemini 3 Pro: 44.0 (#75)
| Benchmark | Claude Sonnet 4 | Gemini 3 Pro |
|---|---|---|
| LMArena Longer Query | 1398 | 1472 |
| Fiction.LiveBench | 46.9% | — |
| CL-bench | — | 15.8% |
Writing & Preference Gemini 3 Pro leads
Claude Sonnet 4: 57.3 (#127), Gemini 3 Pro: 66.6 (#34)
| Benchmark | Claude Sonnet 4 | Gemini 3 Pro |
|---|---|---|
| LMArena Text | 1351 | 1479 |
| LMArena Creative Writing | 1345 | 1482 |
| EQ-Bench Creative Writing | 1483 | 1525 |
| WildBench | 83.8% | 85.9% |
| LMArena Multi-Turn | 1376 | 1484 |
| Short-Story Creative Writing | 81.4% | — |
Frequently asked questions
Is Claude Sonnet 4 better than Gemini 3 Pro?
Gemini 3 Pro is the stronger model overall, scoring 54.7 to 40.6 on the Noometry Index.
Is Claude Sonnet 4 or Gemini 3 Pro better for coding?
Gemini 3 Pro scores higher on coding benchmarks: 52.1 versus 44.0 in the Noometry coding category.
How many benchmarks do Claude Sonnet 4 and Gemini 3 Pro share?
46 benchmarks have published results for both models. Claude Sonnet 4 has 58 scored results on Noometry and Gemini 3 Pro has 67.