AnthropicClaude Sonnet 4
Claude Sonnet 4 is an Anthropic proprietary reasoning model, released 22 May 2025. Its best configuration (thinking reasoning effort) ranks #294 of 785 on the BenchLeader Index at 53.2 ±2.9, in the upper half. The ± is the point: 284 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 maths (61) and lowest in reasoning (45). At $6.00 per million tokens blended it is among the most expensive ranked models. It has been measured at 3 reasoning-effort settings; this summary describes the best-scoring one, and the tabs above switch between them. 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 May 2025
- Overall index53
- Reasoning45
- Coding49
- Agents & tools50
- Maths61
- Knowledge57
- Instruction following54
- Human preference58
- Multimodal50
- Long context60
- Composite51
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 4thinkingthis page | 22 May 2025 | 53.2 | #294 |
| Claude 3.7 Sonnetthinking | 24 Feb 2025 | 51.0 | #353 |
| Claude 3.5 Sonnet | 22 Oct 2024 | 46.3 | #490 |
| Claude 3 Sonnet | 29 Feb 2024 | 38.0 | #721 |
| Claude 37 Sonnetthinking | – | – | – |
| Claude Sonnet 5.0max | – | – | – |
Reasoning-effort configurations
The same model behaves differently depending on how much it is allowed to think. Each row is one setting, scored only on the benchmarks that were run at that setting. “Not stated” collects results from publishers that did not say which setting they used; for a reasoning model that is usually its thinking mode, but we do not assume it. Pick a setting here or at the top of the page to see its price, speed and category scores.
| Effort | Index | Rank | Speed | First answer | Chat reply cost | Categories |
|---|---|---|---|---|---|---|
| no reasoning | 49.4 | #398 | – | – | $0.0057 | Agents & tools 58 · Composite 48 · Instruction following 47 · Knowledge 57 · Long context 46 · Multimodal 45 · Reasoning 43 |
| thinkingbest | 53.2 | #294 | – | – | $0.0057 | Agents & tools 50 · Coding 49 · Composite 51 · Human preference 58 · Instruction following 54 · Knowledge 57 · Long context 60 · Maths 61 · Multimodal 50 · Reasoning 45 |
| not stated | 50.5 | #367 | – | – | $0.0057 | Agents & tools 54 · Coding 51 · Human preference 56 · Instruction following 56 · Knowledge 48 · Long context 36 · Maths 56 · Multimodal 53 · Reasoning 45 |
How its index has moved
8 Sept 2026 to 23 Oct 2026- thinking
- not stated
- no reasoning
The 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 | no reasoning | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|
| GPQA Diamondeffort not stated | – | – | 79.2%#129 | Epoch AI Benchmarking Hub | |
| Humanity's Last Exameffort not stated | – | – | 7.8%#37 | Scale AI / CAIS | |
| SimpleBencheffort not stated | – | – | 45.5%#55 | SimpleBench | |
| Arena Hard Prompts | – | 1433#132 | 1419#150 | Arena (LMArena) | |
| GPQA Diamond (AA)not in index | 68.3%#314 | 77.7%#217 | – | Artificial Analysis | |
| Humanity's Last Exam (AA)not in index | 4.3%#515 | 10.7%#313 | – | Artificial Analysis | |
| GPQA Diamond (Vals)not in index | – | 75.0%#78 | 69.4%#91 | Vals AI | |
| Kagi LLM Benchmark | – | 73.0%#20 | 55.9%#61 | Kagi LLM Benchmark | |
| GPQA Diamond (HELM)not in index | – | 70.6%#10 | 64.3%#22 | HELM Capabilities | |
| ARC-AGI-1 | – | 40.0%#192 | 23.8%#214 | ARC Prize | |
| ARC-AGI-2 | – | 5.9%#172 | 1.3%#218 | ARC Prize | |
| CritPt | 1.1%#223 | 0.3%#271 | – | Artificial Analysis | |
| LMCAnot in indexeffort not stated | – | – | 29.0%#142 | Epoch AI Benchmarking Hub | |
| DTBenchnot in indexeffort not stated | – | – | 77.1%#142 | Epoch AI Benchmarking Hub | |
| ForecastBenchnot in indexeffort not stated | – | – | 64.2%#31 | ForecastBench |
Coding
| Benchmark | thinking | not stated | Source | Trend |
|---|---|---|---|---|
| SciCodeeffort not stated | – | 40.0%#162 | SciCode | |
| WeirdMLeffort not stated | – | 46.1%#80 | WeirdML | |
| GSO-Bencheffort not stated | – | 4.9%#25 | GSO-Bench | |
| Arena Coding | 1474#112 | 1449#147 | Arena (LMArena) | |
| LiveCodeBench | 62.4%#99 | 59.7%#102 | Vals AI | |
| SWE-Bench Proeffort not stated | – | 42.7%#9 | Scale AI SEAL | |
| Aider Polygloteffort not stated | – | 61.3%#13 | Aider polyglot leaderboard | |
| SWE-bench Verified (bash only)effort not stated | – | 64.9%#20 | SWE-bench | |
| SWE-bench Verified (any scaffold)not in indexeffort not stated | – | 76.8%#3 | SWE-bench | |
| SWE-Bench Pro (private)not in indexeffort not stated | – | 42.7%#9 | Scale AI SEAL | |
| ALE-Benchnot in indexeffort not stated | – | 655.4#91 | Epoch AI Benchmarking Hub | |
| Terminal-Bench 2.1 (AA)not in index | 36.3%#146 | – | Artificial Analysis |
Agents & tools
| Benchmark | no reasoning | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|
| Cybencheffort not stated | – | – | 35.0%#8 | Cybench | |
| Terminal-Bench Hard | 27.3%#125 | 31.1%#106 | – | Artificial Analysis | |
| τ²-Bench Telecom (AA)not in index | 52.3%#194 | 64.6%#170 | – | Artificial Analysis | |
| τ³-Banking (AA) | – | 16.7%#109 | – | Artificial Analysis | |
| DeepResearch Benchnot in indexeffort not stated | – | – | 46.6%#27 | DeepResearch Bench | |
| TheAgentCompanynot in indexeffort not stated | – | – | 33.1%#3 | TheAgentCompany | |
| METR Time Horizonsnot in indexeffort not stated | – | – | 62.0%#19 | METR | |
| BrowseComp-Plusnot in indexeffort not stated | – | – | 14.7%#14 | BrowseComp-Plus | |
| BrowseCompnot in indexeffort not stated | – | – | 0.6%#23 | OpenAI |
Maths
| Benchmark | thinking | not stated | Source | Trend |
|---|---|---|---|---|
| OTIS Mock AIMEeffort not stated | – | 71.1%#135 | Epoch AI Benchmarking Hub | |
| MATH Level 5effort not stated | – | 84.4%#28 | Epoch AI Benchmarking Hub | |
| AIME (Vals) | 76.3%#51 | 38.5%#68 | Vals AI | |
| MGSM | 90.9%#31 | 93.0%#11 | Vals AI | |
| Omni-MATH (HELM) | 60.2%#10 | 51.3%#17 | HELM Capabilities | |
| MATH 500not in index | 93.8%#15 | 90.3%#26 | Vals AI | |
| Arena Maths | 1404#143 | 1390#163 | Arena (LMArena) |
Knowledge
| Benchmark | no reasoning | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|
| AA-Omniscience | -9.0#184 | 0.2#136 | – | Artificial Analysis | |
| MMLU-Pro | – | 83.9%#67 | 79.4%#90 | Vals AI | |
| LegalBench | – | 82.1%#74 | 83.0%#62 | Vals AI | |
| CorpFin | – | 61.2%#55 | 54.7%#89 | Vals AI | |
| TaxEval | – | 72.0%#67 | 69.6%#93 | Vals AI | |
| MedQA | – | 92.7%#27 | 90.3%#45 | Vals AI | |
| MultiNRCeffort not stated | – | – | 18.4%#35 | Scale AI SEAL | |
| MMLU-Pro (HELM)not in index | – | 84.3%#10 | 84.3%#10 | HELM Capabilities | |
| MedCodenot in index | – | 35.0%#76 | 33.9%#80 | Vals AI | |
| MedScribenot in index | – | 69.3%#92 | 72.4%#85 | Vals AI | |
| AA-Omniscience: accuracynot in index | 22.7%#270 | 22.7%#269 | – | Artificial Analysis | |
| AA-Omniscience: non-hallucinationnot in index | 59.0%#82 | 70.9%#38 | – | Artificial Analysis |
Instruction following
| Benchmark | no reasoning | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|
| IFBench | 45.4%#205 | 54.7%#150 | – | Artificial Analysis | |
| MultiChallenge | – | 57.1%#19 | – | Scale AI SEAL | |
| IFEval (HELM) | – | 84.0%#25 | 83.9%#27 | HELM Capabilities | |
| WildBench (HELM) | – | 83.8%#16 | 82.5%#23 | HELM Capabilities | |
| Arena Instruction Followingnot in index | – | 1416#111 | 1396#143 | Arena (LMArena) |
Human preference
| Benchmark | thinking | not stated | Source | Trend |
|---|---|---|---|---|
| Arena Text | 1402#148 | 1391#161 | Arena (LMArena) | |
| Arena Creative Writingnot in index | 1396#110 | 1388#125 | Arena (LMArena) | |
| Arena Multi-turnnot in index | 1421#125 | 1410#141 | Arena (LMArena) | |
| Arena Longer Queriesnot in index | 1442#103 | 1427#125 | Arena (LMArena) | |
| Lech Mazur Creative Writingnot in indexeffort not stated | – | 81.4%#14 | Lech Mazur |
Multimodal
| Benchmark | no reasoning | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|
| Arena Vision | – | 1191#97 | 1175#105 | Arena (LMArena) | |
| MMMU-Pro | 62.4%#195 | 61.8%#205 | – | Artificial Analysis | |
| VISTA | – | 45.5%#24 | 43.2%#34 | Scale AI SEAL | |
| MMMU (validation)effort not stated | – | – | 74.4%#16 | MMMU | |
| MMMU-Pro (Vals)not in index | – | 74.9%#55 | 72.4%#62 | Vals AI | |
| MortgageTaxnot in index | – | 49.9%#80 | 62.5%#52 | Vals AI | |
| SAGEnot in index | – | 32.0%#73 | 35.0%#68 | Vals AI | |
| VTBnot in index | – | 4.4%#18 | 4.5%#17 | Scale AI SEAL | |
| GeoBenchnot in indexeffort not stated | – | – | 37.0%#25 | GeoBench | |
| VPCTnot in indexeffort not stated | – | – | 34.0%#27 | Epoch AI Benchmarking Hub | |
| MindCubenot in indexeffort not stated | – | – | 44.8%#3 | Epoch AI Benchmarking Hub |
Long context
| Benchmark | no reasoning | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|
| Fiction.LiveBench 120keffort not stated | – | – | 36.4%#29 | Fiction.live | |
| AA-LCR | 44.0%#342 | 70.3%#211 | – | Artificial Analysis |
Composite
| Benchmark | no reasoning | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|
| Epoch Capabilities Indexnot in indexeffort not stated | – | – | 141.6#106 | Epoch AI Benchmarking Hub | |
| AA Intelligence Index v4.3.2 | 16.6#273 | 18.9#246 | – | Artificial Analysis | |
| HELM Capabilities meannot in index | – | 76.6%#11 | 73.3%#16 | HELM Capabilities |
Safety & honesty
| Benchmark | thinking | not stated | Source | Trend |
|---|---|---|---|---|
| FORTRESSnot in index | 18.1%#43 | 24.4%#33 | Scale AI SEAL | |
| MASKnot in index | 95.3%#3 | 89.3%#9 | Scale AI SEAL | |
| PropensityBenchnot in indexeffort not stated | – | 12.2%#13 | 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 at $0.300 per 1M. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0057 | $0.0049 | – |
| Summarize a 30-page report | 12,000 / 600 | $0.045 | $0.021 | – |
| Code edit | 6,000 / 1,500 | $0.041 | $0.028 | – |
| Agentic coding session | 60,000 / 4,000 | $0.240 | $0.118 | – |
| Structured extraction | 2,000 / 200 | $0.0090 | $0.0050 | – |
More from Anthropic
See also
Questions people ask
- What are Claude Sonnet 4's benchmark scores?
- As of 24 Oct 2026, Claude Sonnet 4 (thinking effort) scores 53.2 ±2.9 on the BenchLeader Index, strongest in maths (61) and weakest in reasoning (45). That rests on 29 independent benchmark results, and every score on this page links to its publisher.
- How does Claude Sonnet 4 rank?
- Claude Sonnet 4 (thinking effort) is #294 of 785 ranked configurations on 24 Oct 2026. 284 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 Sonnet 4 cost per 1M tokens?
- As of 24 Oct 2026, Claude Sonnet 4 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 Sonnet 4 released?
- Anthropic released Claude Sonnet 4 on 22 May 2025. BenchLeader has results for it at 3 reasoning-effort settings.
Data as of 24 Oct 2026. Compare these configurations.
Cite as: BenchLeader, “Claude Sonnet 4: benchmarks, pricing, speed and rank”, data as of 24 Oct 2026, https://www.benchleader.com/models/claude-sonnet-4.