# Claude Sonnet 4.5 vs Gemini 3.1 Pro Preview

> Gemini 3.1 Pro Preview is the stronger model overall, scoring 56.7 to 44.1 on the Noometry Index.

- Canonical page: https://noometry.com/compare/claude-sonnet-4-5-vs-gemini-3-1-pro-preview
- Last updated: 2026-10-12
- Shared benchmarks: 55

## Summary

- They share 55 benchmarks with published results for both. Claude Sonnet 4.5 scores higher in 3 categories and Gemini 3.1 Pro Preview in 7 categories; 8 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Gemini 3.1 Pro Preview leads 71.7 to 26.9.
- The biggest single-benchmark swing is ARC-AGI-2: 13.6% for Claude Sonnet 4.5 and 77.1% for Gemini 3.1 Pro Preview.
- Gemini 3.1 Pro Preview is cheaper at $2 / $12 per million input/output tokens, against $3 / $15 for Claude Sonnet 4.5.
- Gemini 3.1 Pro Preview accepts more context: 1.05M tokens versus 200K.

## Snapshot

| | Claude Sonnet 4.5 | Gemini 3.1 Pro Preview |
|---|---|---|
| Provider | Anthropic | Google |
| Noometry Index | 44.1 | 56.7 |
| Rank | 81 | 23 |
| Context | 200K | 1.05M |
| Input $/M | $3 | $2 |
| Output $/M | $15 | $12 |
| Weights | Proprietary | Proprietary |

## Coding

- Claude Sonnet 4.5: 47.3 (#61)
- Gemini 3.1 Pro Preview: 42.5 (#99)

| Benchmark | Claude Sonnet 4.5 | Gemini 3.1 Pro Preview |
|---|---|---|
| SWE-bench Verified | 71.3% | 75.6% |
| LMArena WebDev | 1393 | 1447 |
| SciCode | 44.7% | 58.9% |
| GSO | 14.7% | 22.6% |
| WeirdML | 47.7% | 72.1% |
| LMArena Coding | 1489 | 1484 |
| ALE-Bench | 796.15 | 1,161 |
| AlgoTune | 1.52 | 2.02 |
| DeepSWE | — | 11.7% |
| SWE-bench Verified (bash only) | 71.4% | — |
| SWE-bench Multilingual | 67% | — |
| MirrorCode | — | 8.9% |

## Agentic & Tool Use

- Claude Sonnet 4.5: 38.3 (#32)
- Gemini 3.1 Pro Preview: 37.7 (#34)

| Benchmark | Claude Sonnet 4.5 | Gemini 3.1 Pro Preview |
|---|---|---|
| Terminal-Bench | 46.5% | 80.2% |
| τ²-bench Banking | 25.3% | 26% |
| DeepResearch Bench | 52.6% | 47.8% |
| LMArena Search | 1159 | 1211 |
| METR Time Horizons | 67.4% | 77% |
| Vending-Bench 2 | 3,839 | 3,774 |
| APEX-Agents | — | 35.3% |
| Berkeley Function Calling Leaderboard | 73.2% | — |
| GDPval | 42.5% | — |
| Remote Labor Index | 2.1% | — |
| τ²-bench Airline | 72% | — |
| τ²-bench Retail | 72.4% | — |
| τ²-bench Telecom | 84.9% | — |
| Cybench | 60% | — |
| OSWorld | 62.9% | — |
| PostTrainBench | — | 22% |
| BALROG | — | 57% |
| ExploitBench | — | 26.1% |
| GBAEval | — | 0.8% |
| GDP.pdf | — | 17% |

## Reasoning

- Claude Sonnet 4.5: 26.9 (#125)
- Gemini 3.1 Pro Preview: 71.7 (#12)

| Benchmark | Claude Sonnet 4.5 | Gemini 3.1 Pro Preview |
|---|---|---|
| ARC-AGI-2 | 13.6% | 77.1% |
| SimpleBench | 54.3% | 79.6% |
| NYT Connections (extended) | 37.3% | 97.4% |
| ARC-AGI-1 | 63.7% | 98% |
| CritPt | 1.1% | 17.7% |
| Chess Puzzles | 12% | 55% |
| EnigmaEval | 6% | 36.8% |
| EBR-Bench | 2.4% | 14.3% |
| LMArena Hard Prompts | 1462 | 1485 |
| Mystery Game Puzzles | 17% | 34% |
| DTBench | 83.2% | 97.1% |
| LMCA | 38.8% | 53.8% |
| Epoch Capabilities Index | 146.84 | 154.77 |
| ForecastBench | 61.9 | 59 |
| Kagi LLM Benchmark | 57.9% | — |
| Thematic Generalization | — | 79.4% |

## Math

- Claude Sonnet 4.5: 32.3 (#216)
- Gemini 3.1 Pro Preview: 62.1 (#34)

| Benchmark | Claude Sonnet 4.5 | Gemini 3.1 Pro Preview |
|---|---|---|
| FrontierMath (Tiers 1-3) | 23.9% | 59.6% |
| FrontierMath Tier 4 | 2.4% | 26.8% |
| OTIS Mock AIME 2024-2025 | 77.8% | 95.6% |
| ProofBench | 19% | 26% |
| LMArena Math | 1449 | 1485 |
| FrontierMath (Feb 2025 set) | 15.2% | 36.9% |
| FrontierMath Tier 4 (v1) | 4.2% | 16.7% |
| MathArena Final-Answer Competitions | — | 86.5% |
| Omni-MATH | 55.3% | — |
| MATH Level 5 | 97.7% | — |

## Knowledge

- Claude Sonnet 4.5: 48.4 (#76)
- Gemini 3.1 Pro Preview: 71.8 (#3)

| Benchmark | Claude Sonnet 4.5 | Gemini 3.1 Pro Preview |
|---|---|---|
| GPQA Diamond | 82.3% | 94.4% |
| Humanity's Last Exam | 13.7% | 46.4% |
| SimpleQA Verified | 30.7% | 73.5% |
| Vectara Hallucination Rate | 12% | 10.4% |
| LMArena Expert | 1482 | 1485 |
| MMLU-Pro | 86.9% | — |
| GPQA (HELM) | 68.6% | — |

## Multimodal

- Claude Sonnet 4.5: 34.8 (#89)
- Gemini 3.1 Pro Preview: 37.9 (#69)

| Benchmark | Claude Sonnet 4.5 | Gemini 3.1 Pro Preview |
|---|---|---|
| LMArena Document | 1450 | 1444 |
| LMArena Vision | — | 1296 |
| VPCT | 39.8% | — |
| Blueprint-Bench 2 | — | 26.5% |
| Furniture Assembly | — | 26.7% |

## Multilingual

- Claude Sonnet 4.5: 53.4 (#69)
- Gemini 3.1 Pro Preview: 57.0 (#12)

| Benchmark | Claude Sonnet 4.5 | Gemini 3.1 Pro Preview |
|---|---|---|
| LMArena Non-English | 1425 | 1477 |
| LMArena Chinese | 1459 | 1529 |
| LMArena French | 1458 | 1487 |
| LMArena German | 1427 | 1491 |
| LMArena Japanese | 1390 | 1493 |
| LMArena Korean | 1403 | 1455 |
| LMArena Russian | 1437 | 1498 |
| LMArena Spanish | 1457 | 1479 |

## Instruction Following

- Claude Sonnet 4.5: 75.0 (#78)
- Gemini 3.1 Pro Preview: 77.0 (#32)

| Benchmark | Claude Sonnet 4.5 | Gemini 3.1 Pro Preview |
|---|---|---|
| LMArena Instruction Following | 1459 | 1466 |
| IFEval | 85% | — |

## Long Context

- Claude Sonnet 4.5: 45.2 (#46)
- Gemini 3.1 Pro Preview: 47.4 (#18)

| Benchmark | Claude Sonnet 4.5 | Gemini 3.1 Pro Preview |
|---|---|---|
| LMArena Longer Query | 1476 | 1483 |
| CL-bench | — | 20.8% |
| CL-bench Life | — | 16.9% |

## Writing & Preference

- Claude Sonnet 4.5: 66.5 (#34)
- Gemini 3.1 Pro Preview: 66.1 (#37)

| Benchmark | Claude Sonnet 4.5 | Gemini 3.1 Pro Preview |
|---|---|---|
| LMArena Text | 1439 | 1481 |
| LMArena Creative Writing | 1442 | 1482 |
| EQ-Bench Creative Writing | 1678 | 1491 |
| LMArena Multi-Turn | 1465 | 1488 |
| WildBench | 85.4% | — |
| EQ-Bench 4 | — | 1142 |

## FAQ

### Is Claude Sonnet 4.5 better than Gemini 3.1 Pro Preview?

Gemini 3.1 Pro Preview is the stronger model overall, scoring 56.7 to 44.1 on the Noometry Index.

### Which is cheaper, Claude Sonnet 4.5 or Gemini 3.1 Pro Preview?

Gemini 3.1 Pro Preview is cheaper. It lists at $2 per million input tokens and $12 per million output tokens; Claude Sonnet 4.5 lists at $3 and $15.

### Is Claude Sonnet 4.5 or Gemini 3.1 Pro Preview better for coding?

Claude Sonnet 4.5 scores higher on coding benchmarks: 47.3 versus 42.5 in the Noometry coding category.

### Which has the bigger context window?

Gemini 3.1 Pro Preview does, with 1.05M tokens against 200K.

### How many benchmarks do Claude Sonnet 4.5 and Gemini 3.1 Pro Preview share?

55 benchmarks have published results for both models. Claude Sonnet 4.5 has 73 scored results on Noometry and Gemini 3.1 Pro Preview has 71.