AnthropicClaude Sonnet 5
Reasoning effort
Claude Sonnet 5 is an Anthropic proprietary reasoning model, released 30 Jun 2026. Its best configuration (high reasoning effort) ranks #114 of 785 on the BenchLeader Index at 60.2 ±1.7, in the upper half. The ± is the point: 200 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 reasoning (66) and lowest in agents & tools (57). At $4.00 per million tokens blended it is among the most expensive ranked models. Output speed of 70 tokens per second puts it slower than most, with a first answer in 7.2 s. It has been measured at 7 reasoning-effort settings; this summary describes the best-scoring one, and the tabs above switch between them. Last measured 22 Oct 2026.
- Blended price
- $4.00/M
- $2.00 in · $10.00 out
- Output speed
- 70 tok/s
- measured by Artificial Analysis
- First answer
- 7.20 s
- first token 1.57 s
- Context
- 1M
- Full answer
- 14 s
- median, reasoning included
- Cost per run
- $1.79
- one full Intelligence Index run
- Released
- 30 Jun 2026
- Overall index60
- Reasoning66
- Coding61
- Agents & tools57
- Maths65
- Knowledge60
- Human preference65
- Multimodal61
- Long context63
- Composite65
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 5highthis page | 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 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 | 51.9 | #336 | 69 tok/s | 0.82 s | $0.0038 | Agents & tools 44 · Composite 55 · Knowledge 61 · Long context 60 · Multimodal 55 · Reasoning 43 |
| low | 53.2 | #298 | 69 tok/s | 1.00 s | $0.0038 | Coding 54 · Composite 57 · Knowledge 58 · Long context 59 · Reasoning 48 |
| medium | 55.3 | #234 | 70 tok/s | 1.71 s | $0.0038 | Coding 56 · Composite 61 · Knowledge 58 · Long context 62 · Reasoning 55 |
| highbest | 60.2 | #114 | 70 tok/s | 7.20 s | $0.0038 | Agents & tools 57 · Coding 61 · Composite 65 · Human preference 65 · Knowledge 60 · Long context 63 · Maths 65 · Multimodal 61 · Reasoning 66 |
| xhigh | 59.0 | #141 | 71 tok/s | 14 s | $0.0038 | Coding 60 · Composite 68 · Knowledge 53 · Long context 63 · Maths 65 · Reasoning 66 |
| max | 58.4 | #153 | 84 tok/s | 136 s | $0.0038 | Agents & tools 46 · Coding 59 · Composite 72 · Knowledge 56 · Long context 66 · Maths 58 · Multimodal 60 · Reasoning 63 |
| not stated | 56.1 | #213 | 73 tok/s | 1.57 s | $0.0038 | Agents & tools 58 · Coding 55 · Composite 56 · Human preference 56 · Knowledge 61 · Reasoning 59 |
What more thinking costs
Turning the effort up buys index points and multiplies the bill. Here max costs 10× low for +5.2 on the index — 1.6 points per doubling of spend. Compare that with other models.
How its index has moved
8 Sept 2026 to 23 Oct 2026- high
- xhigh
- max
- not stated
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 | low | medium | high | xhigh | max | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| GPQA Diamond | – | – | – | – | 90.5%#42 | 80.3%#124 | – | Epoch AI Benchmarking Hub | |
| SimpleBencheffort not stated | – | – | – | – | – | – | 60.6%#29 | SimpleBench | |
| Arena Hard Prompts | – | – | – | 1490#49 | – | – | – | Arena (LMArena) | |
| LiveBench 2.0 Reasoningnot in indexeffort not stated | – | – | – | – | – | – | 88.7%#21 | LiveBench | |
| LiveBench 1.0 Reasoningnot in index | – | – | – | – | 88.7%#20 | – | – | LiveBench | |
| GPQA Diamond (AA)not in index | 80.0%#193 | – | – | – | – | 91.1%#50 | – | Artificial Analysis | |
| Humanity's Last Exam (AA)not in index | 19.1%#225 | 21.9%#205 | 30.0%#151 | 35.7%#114 | 39.0%#94 | 41.3%#79 | – | Artificial Analysis | |
| GPQA Diamond (Vals)not in indexeffort not stated | – | – | – | – | – | – | 88.9%#31 | Vals AI | |
| CritPt | 1.1%#223 | 4.6%#161 | 8.6%#134 | 15.1%#93 | 15.4%#89 | 16.9%#82 | – | Artificial Analysis | |
| Chess Puzzlesnot in index | – | – | – | – | 34.0%#58 | 45.0%#29 | – | Epoch AI Benchmarking Hub | |
| Mystery Game Puzzlesnot in index | – | – | – | – | – | 35.0%#23 | 16.0%#84 | Epoch AI Benchmarking Hub | |
| LMCAnot in index | – | 49.6%#67 | 49.7%#66 | 50.0%#64 | 48.8%#73 | 49.3%#70 | – | Epoch AI Benchmarking Hub | |
| DTBenchnot in index | – | 73.6%#153 | 79.2%#136 | 84.5%#107 | 89.5%#91 | 92.5%#69 | – | Epoch AI Benchmarking Hub | |
| ForecastBenchnot in indexeffort not stated | – | – | – | – | – | – | 63.4%#40 | ForecastBench | |
| BTF-3not in index | – | – | – | – | 14.2%#4 | – | – | Epoch AI Benchmarking Hub |
Coding
| Benchmark | no reasoning | low | medium | high | xhigh | max | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| SciCode | – | 50.1%#109 | 51.6%#93 | 54.3%#62 | 54.0%#67 | 53.6%#72 | – | SciCode | |
| WeirdML | – | – | – | 68.8%#35 | – | – | – | WeirdML | |
| FrontierCodeeffort not stated | – | – | – | – | – | – | 42.7%#20 | Cognition | |
| GSO-Bencheffort not stated | – | – | – | – | – | – | 37.3%#10 | GSO-Bench | |
| Arena Coding | – | – | – | 1520#37 | – | – | – | Arena (LMArena) | |
| Arena WebDev | – | – | – | 1541#45 | – | – | – | Arena (LMArena) | |
| LiveBench 2.0 Codingnot in indexeffort not stated | – | – | – | – | – | – | 74.7%#36 | LiveBench | |
| LiveBench 1.0 Codingnot in index | – | – | – | – | 80.7%#18 | – | – | LiveBench | |
| SciCode (AA)not in index | – | 50.1%#110 | 51.6%#93 | 54.3%#65 | 54.0%#70 | 54.3%#65 | – | Artificial Analysis | |
| LiveCodeBencheffort not stated | – | – | – | – | – | – | 82.4%#51 | Vals AI | |
| IOIeffort not stated | – | – | – | – | – | – | 45.0%#31 | Vals AI | |
| SWE-bench (Vals)not in indexeffort not stated | – | – | – | – | – | – | 79.6%#25 | Vals AI | |
| Code Migrationnot in indexeffort not stated | – | – | – | – | – | – | 44.4%#22 | Vals AI | |
| ProgramBenchnot in indexeffort not stated | – | – | – | – | – | – | 0.0%#32 | Vals AI | |
| Vibe Code Bench 1-100not in indexeffort not stated | – | – | – | – | – | – | 13.8%#22 | Vals AI | |
| Vibe Code Bench v1.1not in indexeffort not stated | – | – | – | – | – | – | 81.3%#24 | Vals AI | |
| Surface Evolver Benchnot in index | – | – | 60.0%#13 | – | – | – | – | Epoch AI Benchmarking Hub | |
| DeepSWE v1.1not in index | – | 30.5%#63 | 39.8%#57 | 48.2%#50 | 49.7%#48 | 53.9%#41 | – | Epoch AI Benchmarking Hub | |
| CursorBenchnot in index | – | 24.1%#76 | 28.0%#69 | 30.8%#65 | 32.0%#61 | 34.1%#53 | – | Cursor | |
| ALE-Benchnot in index | – | – | – | 1463.1#25 | – | – | – | Epoch AI Benchmarking Hub | |
| Terminal-Bench 4.0 (AA)not in index | – | 2.5%#113 | 2.0%#118 | 5.0%#102 | 7.1%#95 | 14.1%#74 | – | Artificial Analysis | |
| Terminal-Bench 2.1 (AA)not in index | 75.3%#62 | – | – | – | – | 80.5%#42 | – | Artificial Analysis | |
| Startrise Frontendnot in indexeffort not stated | – | – | – | – | – | – | 66.6%#23 | Startrise |
Agents & tools
| Benchmark | no reasoning | low | medium | high | xhigh | max | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| Terminal-Bench | – | – | – | – | – | 12.4%#79 | – | Terminal-Bench | |
| APEX-Agentseffort not stated | – | – | – | – | – | – | 54.5%#21 | Mercor | |
| Arena Agent | – | – | – | 4.2#11 | – | – | – | Arena (LMArena) | |
| LiveBench Long Agenticnot in indexeffort not stated | – | – | – | – | – | – | 39.0%#26 | LiveBench | |
| LiveBench Browsernot in indexeffort not stated | – | – | – | – | – | – | 72.0%#21 | LiveBench | |
| LiveBench Agentic Codingnot in index | – | – | – | – | 59.4%#17 | – | – | LiveBench | |
| Terminal-Bench 2.1 (Vals)effort not stated | – | – | – | – | – | – | 74.5%#18 | Vals AI | |
| τ³-Banking (AA) | 15.7%#114 | – | – | – | – | 37.3%#41 | – | Artificial Analysis | |
| Analyst Agent (AA)not in index | – | – | – | – | – | 46.3%#15 | – | Artificial Analysis | |
| CyberBenchnot in indexeffort not stated | – | – | – | – | – | – | 61.9%#33 | Vals AI | |
| Finance Agent v2not in indexeffort not stated | – | – | – | – | – | – | 53.9%#30 | Vals AI | |
| Harvey's Legal Agent Benchmark v1not in indexeffort not stated | – | – | – | – | – | – | 5.0%#37 | Vals AI | |
| Legal Research Benchnot in indexeffort not stated | – | – | – | – | – | – | 41.8%#24 | Vals AI | |
| SkillsBenchnot in indexeffort not stated | – | – | – | – | – | – | 46.5%#25 | Vals AI | |
| SREBenchnot in indexeffort not stated | – | – | – | – | – | – | 0.4%#29 | Vals AI | |
| Tax Agent Benchnot in indexeffort not stated | – | – | – | – | – | – | 29.4%#30 | Vals AI | |
| Terminal-Bench 4.0 (Vals)not in indexeffort not stated | – | – | – | – | – | – | 10.6%#38 | Vals AI | |
| Terminal-Bench Sciencenot in indexeffort not stated | – | – | – | – | – | – | 5.7%#21 | Vals AI | |
| GBAEvalnot in indexeffort not stated | – | – | – | – | – | – | 65.3%#5 | Epoch AI Benchmarking Hub | |
| Vending-Bench 2not in indexeffort not stated | – | – | – | – | – | – | 6377.7#19 | Andon Labs | |
| AutomationBenchnot in index | – | 19.9%#138 | 27.9%#123 | 32.1%#115 | 34.5%#111 | 36.5%#103 | – | Artificial Analysis | |
| GDP.pdf (AA)not in index | – | 9.4%#135 | 11.6%#117 | 9.4%#135 | 12.6%#111 | 13.2%#107 | – | Artificial Analysis | |
| EnterpriseOps-Gymnot in index | – | – | – | – | 43.9%#19 | 44.7%#15 | – | Artificial Analysis |
Maths
| Benchmark | high | xhigh | max | Source | Trend |
|---|---|---|---|---|---|
| FrontierMath Tiers 1–3 | – | – | 65.6%#36 | Epoch AI Benchmarking Hub | |
| FrontierMath Tier 4 | – | – | 29.3%#40 | Epoch AI Benchmarking Hub | |
| OTIS Mock AIME | – | 94.7%#53 | 80.0%#114 | Epoch AI Benchmarking Hub | |
| ProofBench | – | – | 77.0%#14 | Vals AI | |
| LiveBench Mathematicsnot in index | – | 92.9%#25 | – | LiveBench | |
| Arena Maths | 1473#47 | – | – | Arena (LMArena) |
Knowledge
| Benchmark | no reasoning | low | medium | high | xhigh | max | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| SimpleQA Verified | – | – | – | – | 32.9%#60 | 33.7%#57 | – | Epoch AI Benchmarking Hub | |
| LiveBench 2.0 Data Analysisnot in indexeffort not stated | – | – | – | – | – | – | 71.7%#34 | LiveBench | |
| LiveBench Researchnot in indexeffort not stated | – | – | – | – | – | – | 82.8%#13 | LiveBench | |
| LiveBench 1.0 Data Analysisnot in index | – | – | – | – | 71.7%#47 | – | – | LiveBench | |
| AA-Omniscience | -0.7#140 | -8.2#178 | -6.9#173 | -3.7#154 | 3.2#118 | 16.4#78 | – | Artificial Analysis | |
| MMLU-Proeffort not stated | – | – | – | – | – | – | 87.5%#26 | Vals AI | |
| LegalBencheffort not stated | – | – | – | – | – | – | 83.9%#46 | Vals AI | |
| CorpFineffort not stated | – | – | – | – | – | – | 67.0%#14 | Vals AI | |
| TaxEvaleffort not stated | – | – | – | – | – | – | 75.6%#14 | Vals AI | |
| Excel Modeling Benchmarknot in indexeffort not stated | – | – | – | – | – | – | 66.3%#22 | Vals AI | |
| MedCodenot in indexeffort not stated | – | – | – | – | – | – | 47.5%#33 | Vals AI | |
| MedScribenot in indexeffort not stated | – | – | – | – | – | – | 76.0%#76 | Vals AI | |
| Public Benefits Bench v1.1not in indexeffort not stated | – | – | – | – | – | – | 66.0%#19 | Vals AI | |
| DrugDiscoveryBenchnot in index | – | – | – | – | – | 50.0%#7 | – | Scale AI SEAL | |
| AA-Omniscience: accuracynot in index | 33.8%#163 | 37.4%#144 | 37.1%#145 | 37.4%#142 | 39.0%#131 | 40.0%#124 | – | Artificial Analysis | |
| AA-Omniscience: non-hallucinationnot in index | 48.0%#131 | 27.2%#230 | 30.1%#223 | 34.3%#194 | 41.4%#154 | 60.6%#74 | – | Artificial Analysis |
Instruction following
| Benchmark | high | xhigh | Source | Trend |
|---|---|---|---|---|
| LiveBench Languagenot in index | – | 75.0%#49 | LiveBench | |
| LiveBench Instruction Followingnot in index | – | 63.9%#48 | LiveBench | |
| Arena Instruction Followingnot in index | 1466#41 | – | Arena (LMArena) |
Human preference
| Benchmark | high | not stated | Source | Trend |
|---|---|---|---|---|
| Arena Text | 1462#58 | – | Arena (LMArena) | |
| EQ-Bench 4effort not stated | – | 1236#10 | EQ-Bench | |
| Arena Creative Writingnot in index | 1435#63 | – | Arena (LMArena) | |
| Arena Multi-turnnot in index | 1473#54 | – | Arena (LMArena) | |
| Arena Longer Queriesnot in index | 1481#44 | – | Arena (LMArena) |
Multimodal
| Benchmark | no reasoning | high | max | not stated | Source | Trend |
|---|---|---|---|---|---|---|
| Arena Vision | – | 1273#41 | – | – | Arena (LMArena) | |
| LiveBench Designnot in indexeffort not stated | – | – | – | 82.0%#11 | LiveBench | |
| MMMU-Pro | 71.9%#134 | – | 77.3%#83 | – | Artificial Analysis | |
| MMMU-Pro (Vals)not in indexeffort not stated | – | – | – | 83.0%#35 | Vals AI | |
| MortgageTaxnot in indexeffort not stated | – | – | – | 70.0%#4 | Vals AI | |
| SAGEnot in indexeffort not stated | – | – | – | 48.9%#24 | Vals AI | |
| Blueprint-Bench 2not in indexeffort not stated | – | – | – | 24.9%#23 | Epoch AI Benchmarking Hub |
Long context
| Benchmark | no reasoning | low | medium | high | xhigh | max | Source | Trend |
|---|---|---|---|---|---|---|---|---|
| AA-LCR | 70.0%#215 | 67.3%#240 | 73.7%#173 | 76.7%#140 | 76.7%#140 | 82.0%#47 | Artificial Analysis | |
| Arena Documentnot in index | – | – | – | 1476#15 | – | – | Arena (LMArena) | |
| MLCRnot in index | – | 6.1%#138 | 5.0%#144 | 4.4%#151 | 5.6%#139 | 5.0%#144 | Artificial Analysis |
Composite
| Benchmark | no reasoning | low | medium | high | xhigh | max | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| Epoch Capabilities Indexnot in indexeffort not stated | – | – | – | – | – | – | 156.9#25 | Epoch AI Benchmarking Hub | |
| LiveBench 2.0effort not stated | – | – | – | – | – | – | 73.0%#23 | LiveBench | |
| LiveBench 1.0not in index | – | – | – | – | 76.0%#33 | – | – | LiveBench | |
| AA Intelligence Index v4.3.2 | 23.2#189 | 24.3#177 | 28.1#138 | 31.7#114 | 34.4#94 | 38.2#78 | – | Artificial Analysis | |
| Vals Multimodal Indexnot in indexeffort not stated | – | – | – | – | – | – | 68.8%#8 | Vals AI |
Where to run it
Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.
| Provider | Speed | First token | Input $/M | Output $/M | Context | Quantisation |
|---|---|---|---|---|---|---|
| Google Vertex | 98 tok/s | 2.54 s | $2.20 | $11.00 | 1M | – |
| Google Vertex | 86 tok/s | 0.56 s | $2.20 | $11.00 | 1M | – |
| Anthropic | 79 tok/s | 1.17 s | $2.00 | $10.00 | 1M | – |
| Amazon Bedrock | 77 tok/s | 2.98 s | $2.00 | $10.00 | 1M | – |
| Azure | 76 tok/s | 2.88 s | $2.00 | $10.00 | 1M | – |
| Google Vertex | 72 tok/s | 1.16 s | $2.00 | $10.00 | 1M | – |
| Claude Platform on AWS | 66 tok/s | 2.60 s | $2.00 | $10.00 | 1M | – |
| Amazon Bedrock | 61 tok/s | 3.88 s | $2.20 | $11.00 | 1M | – |
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 $0.200 per 1M. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0038 | $0.0033 | 11.5 s |
| Summarize a 30-page report | 12,000 / 600 | $0.030 | $0.014 | 15.8 s |
| Code edit | 6,000 / 1,500 | $0.027 | $0.019 | 28.7 s |
| Agentic coding session | 60,000 / 4,000 | $0.160 | $0.079 | 1.1 min |
| Structured extraction | 2,000 / 200 | $0.0060 | $0.0033 | 10.1 s |
More from Anthropic
Compare Claude Sonnet 5 with…
- Claude Sonnet 5 vs Claude Opus 5.5
- Claude Sonnet 5 vs Gemini 4 Argon
- Claude Sonnet 5 vs Muse Wonder Full
- Claude Sonnet 5 vs Claude Fable 5.1
- Claude Sonnet 5 vs GPT-6 Astra
- Claude Sonnet 5 vs GPT-6.1 Sol
- Claude Sonnet 5 vs Claude Fable 5
- Claude Sonnet 5 vs Claude Sonnet 5.5
- Claude Sonnet 5 vs Claude Opus 5
- Claude Sonnet 5 vs GPT-5.6 Sol
See also
Questions people ask
- What are Claude Sonnet 5's benchmark scores?
- As of 24 Oct 2026, Claude Sonnet 5 (high effort) scores 60.2 ±1.7 on the BenchLeader Index, strongest in reasoning (66) and weakest in agents & tools (57). That rests on 13 independent benchmark results, and every score on this page links to its publisher.
- How does Claude Sonnet 5 rank?
- Claude Sonnet 5 (high effort) is #114 of 785 ranked configurations on 24 Oct 2026. 200 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 5 cost per 1M tokens?
- As of 24 Oct 2026, Claude Sonnet 5 costs $2.00 per 1M input tokens and $10.00 per 1M output tokens, $4.00 blended (three input tokens per output token).
- Is Claude Sonnet 5 better than GPT-5.6 Sol?
- As of 24 Oct 2026, GPT-5.6 Sol (max) leads on quality (68.0 vs 60.2 on the BenchLeader Index), but Claude Sonnet 5 (high) costs 2.0× less, so it is the better buy unless you need the extra quality. The head-to-head page has the benchmark-by-benchmark detail.
- When was Claude Sonnet 5 released?
- Anthropic released Claude Sonnet 5 on 30 Jun 2026. BenchLeader has results for it at 7 reasoning-effort settings.
Data as of 24 Oct 2026. Compare these configurations.
Cite as: BenchLeader, “Claude Sonnet 5: benchmarks, pricing, speed and rank”, data as of 24 Oct 2026, https://www.benchleader.com/models/claude-sonnet-5.