AnthropicClaude 3.5 Sonnet
Claude 3.5 Sonnet is an Anthropic proprietary model, released 22 Oct 2024. Its best configuration ranks #490 of 785 on the BenchLeader Index at 46.3 ±3.3, in the lower half. The ± is the point: 285 other configurations score within that range, so they and this one cannot be told apart on quality alone — price and speed are what separate them. It scores highest in instruction following (56) and lowest in composite (38). At $6.00 per million tokens blended it is among the most expensive ranked models. Last measured 22 Oct 2026.
- Blended price
- $6.00/M
- $3.00 in · $15.00 out
- Output speed
- –
- First answer
- –
- Context
- 200k
- Full answer
- –
- Cost per run
- –
- Released
- 22 Oct 2024
- Overall index46
- Reasoning42
- Coding44
- Agents & tools43
- Maths43
- Knowledge47
- Instruction following56
- Human preference54
- Multimodal45
- Composite38
Versions
Anthropic has shipped 10 models under this name. Each is ranked on its own results; a newer version often has fewer results so far, which holds its index nearer the average until more arrive.
| Model | Released | Index | Rank |
|---|---|---|---|
| Claude Sonnet 5.5max | 28 Sept 2026 | 69.3 | #13 |
| Claude Sonnet 5high | 30 Jun 2026 | 60.2 | #114 |
| Claude Sonnet 4.6 | 17 Feb 2026 | 57.1 | #185 |
| Claude Sonnet 4.5high | 29 Sept 2025 | 58.1 | #160 |
| Claude Sonnet 4thinking | 22 May 2025 | 53.2 | #294 |
| Claude 3.7 Sonnetthinking | 24 Feb 2025 | 51.0 | #353 |
| Claude 3.5 Sonnetthis page | 22 Oct 2024 | 46.3 | #490 |
| Claude 3 Sonnet | 29 Feb 2024 | 38.0 | #721 |
| Claude 37 Sonnetthinking | – | – | – |
| Claude Sonnet 5.0max | – | – | – |
How its index has moved
8 Sept 2026 to 18 Oct 2026The index is recomputed from scratch every day, so a line moves when a new benchmark result lands, when a publisher revises a score, or when the models it is normalised against change. Early movement usually means the score is still settling.
Benchmark results
See all LLM benchmarks. One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.
Reasoning
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| GPQA Diamond | 55.3%#205 | Epoch AI Benchmarking Hub | |
| Humanity's Last Exam | 4.1%#46 | Scale AI / CAIS | |
| SimpleBench | 41.4%#57 | SimpleBench | |
| Arena Hard Prompts | 1399#175 | Arena (LMArena) | |
| GPQA Diamond (AA)not in index | 59.9%#377 | Artificial Analysis | |
| Humanity's Last Exam (AA)not in index | 3.7%#576 | Artificial Analysis | |
| GPQA Diamond (Vals)not in index | 59.3%#105 | Vals AI | |
| GPQA Diamond (HELM)not in index | 56.5%#30 | HELM Capabilities | |
| DTBenchnot in index | 67.8%#163 | Epoch AI Benchmarking Hub | |
| ForecastBenchnot in index | 63.2%#46 | ForecastBench |
Coding
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| WeirdML | 40.0%#103 | WeirdML | |
| GSO-Bench | 4.6%#28 | GSO-Bench | |
| Arena Coding | 1436#161 | Arena (LMArena) | |
| LiveCodeBench | 49.6%#112 | Vals AI | |
| Aider Polyglot | 51.6%#25 | Aider polyglot leaderboard | |
| SWE-bench Verified (any scaffold)not in index | 63.4%#25 | SWE-bench | |
| CadEvalnot in index | 48.0%#7 | Epoch AI Benchmarking Hub |
Agents & tools
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Cybench | 17.5%#12 | Cybench | |
| BALROGnot in index | 32.6%#15 | BALROG | |
| TheAgentCompanynot in index | 24.0%#6 | TheAgentCompany | |
| METR Time Horizonsnot in index | 45.2%#28 | METR | |
| BrowseCompnot in index | 0.9%#19 | OpenAI |
Maths
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| OTIS Mock AIME | 8.5%#230 | Epoch AI Benchmarking Hub | |
| MATH Level 5 | 57.0%#48 | Epoch AI Benchmarking Hub | |
| AIME (Vals) | 10.0%#85 | Vals AI | |
| MGSM | 92.6%#15 | Vals AI | |
| Omni-MATH (HELM) | 27.6%#49 | HELM Capabilities | |
| MATH 500not in index | 72.4%#47 | Vals AI | |
| Arena Maths | 1352#198 | Arena (LMArena) |
Knowledge
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| MMLU-Pro | 78.4%#98 | Vals AI | |
| CorpFin | 53.6%#94 | Vals AI | |
| TaxEval | 70.2%#90 | Vals AI | |
| MedQA | 83.2%#61 | Vals AI | |
| MMLU-Pro (HELM)not in index | 77.7%#27 | HELM Capabilities | |
| MMLUnot in index | 87.3%#2 | Epoch AI Benchmarking Hub |
Instruction following
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| IFEval (HELM) | 85.6%#18 | HELM Capabilities | |
| WildBench (HELM) | 79.2%#37 | HELM Capabilities | |
| Arena Instruction Followingnot in index | 1374#168 | Arena (LMArena) |
Human preference
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Arena Text | 1374#179 | Arena (LMArena) | |
| Arena Creative Writingnot in index | 1371#146 | Arena (LMArena) | |
| Arena Multi-turnnot in index | 1390#167 | Arena (LMArena) | |
| Arena Longer Queriesnot in index | 1396#164 | Arena (LMArena) | |
| Lech Mazur Creative Writingnot in index | 80.3%#16 | Lech Mazur |
Multimodal
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Arena Vision | 1126#122 | Arena (LMArena) | |
| VISTA | 38.7%#40 | Scale AI SEAL | |
| MMMU (validation) | 68.3%#24 | MMMU | |
| MMMU-Pro (official)not in index | 51.5%#16 | MMMU | |
| MMMU-Pro (Vals)not in index | 68.8%#70 | Vals AI | |
| MortgageTaxnot in index | 64.1%#47 | Vals AI | |
| GeoBenchnot in index | 62.0%#17 | GeoBench | |
| VPCTnot in index | 33.0%#29 | Epoch AI Benchmarking Hub | |
| Video-MMEnot in index | 60.0%#26 | Video-MME |
Composite
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| Epoch Capabilities Indexnot in index | 133.4#138 | Epoch AI Benchmarking Hub | |
| AA Intelligence Index v4.3.2 | 7.9#452 | Artificial Analysis | |
| HELM Capabilities meannot in index | 65.3%#34 | HELM Capabilities |
Safety & honesty
| Benchmark | not stated | Source | Trend |
|---|---|---|---|
| FORTRESSnot in index | 13.0%#61 | Scale AI SEAL | |
| MASKnot in index | 72.3%#29 | Scale AI SEAL |
What a task costs
Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0057 | – | – |
| Summarize a 30-page report | 12,000 / 600 | $0.045 | – | – |
| Code edit | 6,000 / 1,500 | $0.041 | – | – |
| Agentic coding session | 60,000 / 4,000 | $0.240 | – | – |
| Structured extraction | 2,000 / 200 | $0.0090 | – | – |
More from Anthropic
See also
Questions people ask
- What are Claude 3.5 Sonnet's benchmark scores?
- As of 24 Oct 2026, Claude 3.5 Sonnet scores 46.3 ±3.3 on the BenchLeader Index, strongest in instruction following (56) and weakest in composite (38). That rests on 27 independent benchmark results, and every score on this page links to its publisher.
- How does Claude 3.5 Sonnet rank?
- Claude 3.5 Sonnet is #490 of 785 ranked configurations on 24 Oct 2026. 285 other configurations score within its margin, so the exact position matters less than the band. The leader is Claude Opus 5.5 (max effort) at 72.6.
- How much does Claude 3.5 Sonnet cost per 1M tokens?
- As of 24 Oct 2026, Claude 3.5 Sonnet costs $3.00 per 1M input tokens and $15.00 per 1M output tokens, $6.00 blended (three input tokens per output token).
- When was Claude 3.5 Sonnet released?
- Anthropic released Claude 3.5 Sonnet on 22 Oct 2024.
Data as of 24 Oct 2026. Compare with another model.
Cite as: BenchLeader, “Claude 3.5 Sonnet: benchmarks, pricing, speed and rank”, data as of 24 Oct 2026, https://www.benchleader.com/models/claude-3-5-sonnet.