Across 13 shared benchmarks, Claude Sonnet 4.5 leads overall: Claude Sonnet 4.5 wins 11, Claude Sonnet 3.7 wins 2, with 0 ties and an average score difference of +10.74.
Anthropic · 2025-09-30 · AI model
Anthropic · 2025-02-25 · AI model
Grouped by capability, sorted by largest gap within each. 13 shared benchmarks.
| Benchmark | Claude Sonnet 4.5 | Claude Sonnet 3.7 | Diff |
|---|---|---|---|
| τ²-Bench - Telecom | 985 / 35thinking + 使用工具 | 5531 / 35thinking + 使用工具 | +43 |
| Terminal Bench Hard | 338 / 13thinking + 使用工具 | 2113 / 13thinking + 使用工具 | +12 |
| τ²-Bench | 7124 / 40normal + 使用工具 | 61.8029 / 40thinking + 使用工具 | +9.20 |
| Benchmark | Claude Sonnet 4.5 | Claude Sonnet 3.7 | Diff |
|---|---|---|---|
| LiveBench | 70.5620 / 52 | 60.4043 / 52 | +10.16 |
| GPQA Diamond | 73.7094 / 175 | 68119 / 175 | +5.70 |
| HLE | 7.10136 / 149 | 10.30123 / 149thinking | -3.20 |
| Benchmark | Claude Sonnet 4.5 | Claude Sonnet 3.7 | Diff |
|---|---|---|---|
| AIME2025 | 3796 / 106 | 54.8084 / 106 | -17.80 |
| Simple Bench | 54.309 / 27 | 44.9016 / 27 | +9.40 |
| FrontierMath | 5.2038 / 60 | 3.1046 / 60 | +2.10 |
| Benchmark | Claude Sonnet 4.5 | Claude Sonnet 3.7 | Diff |
|---|---|---|---|
| OSWorld-Verified | 61.4010 / 14thinking + 使用工具 | 2814 / 14thinking + 使用工具 | +33.40 |
| Benchmark | Claude Sonnet 4.5 | Claude Sonnet 3.7 | Diff |
|---|---|---|---|
| SWE-bench Verified | 823 / 103parallel_thinking + 使用工具 | 62.3069 / 103normal + 使用工具 | +19.70 |
| Benchmark | Claude Sonnet 4.5 | Claude Sonnet 3.7 | Diff |
|---|---|---|---|
| AA-LCR | 668 / 13thinking | 6113 / 13thinking | +5 |
| Benchmark | Claude Sonnet 4.5 | Claude Sonnet 3.7 | Diff |
|---|---|---|---|
| GDPval-AA | 3915 / 20thinking | 2819 / 20thinking | +11 |
| Field | Claude Sonnet 4.5 | Claude Sonnet 3.7 |
|---|---|---|
| Publisher | Anthropic | Anthropic |
| Release date | 2025-09-30 | 2025-02-25 |
| Model type | AI model | AI model |
| Architecture | Dense | Dense |
| Parameters | 0.0 | Not available |
| Context length | 1000K | 128K |
| Max output | 65536 | Not available |
Prices use DataLearner records when available; missing fields are not inferred.
| Item | Claude Sonnet 4.5 | Claude Sonnet 3.7 |
|---|---|---|
| Text input | 3 美元/100 万tokens | Not public |
| Text output | 15 美元/100 万tokens | Not public |
| Cache read | 3.75 美元/100 万tokens | Not public |
| Cache write | 0.3 美元/100 万tokens | Not public |
One or both models have incomplete public pricing.
On average across the 13 shared benchmarks, Claude Sonnet 4.5 scores 10.74 higher.
Largest single-benchmark gap: τ²-Bench - Telecom — Claude Sonnet 4.5 98 vs Claude Sonnet 3.7 55 (+43).
Page generated from structured model, pricing and benchmark records. No real-time LLM is used to write the prose.
