AnthropicReasoning modelClaude Opus 4.6
Claude Opus 4.6 is an Anthropic proprietary reasoning model, released 17 Dec 2025. Its best configuration ranks #68 of 756 on the BenchLeader Index at 63.1 ±4.7, in the upper half. The ± is the point: 152 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 agents & tools (69) and lowest in coding (60). At $10.00 per million tokens blended it is among the most expensive ranked models. Output speed of 38 tokens per second puts it in the slowest quarter, with a first token in 1.5 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 2 Oct 2026.
- Blended price
- $10.00/M
- $5.00 in · $25.00 out
- Output speed
- 38 tok/s
- OpenRouter traffic, 7-day median; not yet measured by Artificial Analysis
- First answer
- 1.54 s
- Context
- 1M
- Full answer
- –
- Cost per run
- –
- Released
- 17 Dec 2025
- Overall index63
- Reasoning62
- Coding60
- Agents & tools69
- Maths68
- Human preference64
- Multimodal67
Versions
Anthropic has shipped 9 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 Opus 5.5high | 22 Sept 2026 | 70.1 | #2 |
| Claude Opus 5high | 24 Jul 2026 | 68.4 | #14 |
| Claude Opus 4.8max | 28 May 2026 | 62.4 | #82 |
| Claude Opus 4.7 | 16 Apr 2026 | 63.8 | #59 |
| Claude Opus 4.6this page | 17 Dec 2025 | 63.1 | #68 |
| Claude Opus 4.5thinking | 24 Nov 2025 | 60.2 | #106 |
| Claude Opus 4.1thinking | 5 Aug 2025 | 56.7 | #181 |
| Claude Opus 4thinking | 22 May 2025 | 55.3 | #214 |
| Claude 3 Opus | 4 Mar 2024 | 39.3 | #664 |
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 | 54.3 | #241 | 38 tok/s | 1.54 s | $0.0095 | Instruction following 50 · Knowledge 65 · Multimodal 56 |
| low | – | – | 38 tok/s | 1.54 s | $0.0095 | Reasoning 61 |
| medium | – | – | 38 tok/s | 1.54 s | $0.0095 | Reasoning 63 |
| high | 60.1 | #109 | 38 tok/s | 1.97 s | $0.0095 | Agents & tools 76 · Coding 63 · Composite 54 · Human preference 71 · Instruction following 47 · Knowledge 63 · Long context 59 · Maths 64 · Multimodal 61 · Reasoning 59 |
| max | 57.2 | #170 | 41 tok/s | 20 s | $0.0095 | Agents & tools 58 · Composite 66 · Instruction following 38 · Knowledge 63 · Long context 64 · Maths 57 · Multimodal 58 · Reasoning 62 |
| thinking | 62.5 | #81 | 38 tok/s | 1.54 s | $0.0095 | Coding 64 · Knowledge 62 · Maths 65 · Reasoning 76 |
| not statedbest | 63.1 | #68 | 38 tok/s | 1.54 s | $0.0095 | Agents & tools 69 · Coding 60 · Human preference 64 · Maths 68 · Multimodal 67 · Reasoning 62 |
How its index has moved
8 Sept 2026 to 5 Oct 2026- not stated
- high
- max
- 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
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
Coding
Agents & tools
| Benchmark | no reasoning | low | medium | high | max | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| Terminal-Bench | – | – | – | – | – | – | 79.8%#5 | Terminal-Bench | |
| Cybench | – | – | – | – | – | – | 93.0%#1 | Cybench | |
| Remote Labor Index | – | – | – | – | – | – | 4.2%#8 | Scale AI / CAIS | |
| APEX-Agents | – | – | – | – | 46.3%#33 | – | – | Mercor | |
| LiveBench Agentic Codingnot in index | – | – | – | 49.0%#42 | – | – | – | LiveBench | |
| Terminal-Bench Hard | – | – | – | 48.5%#26 | 46.2%#30 | – | – | Artificial Analysis | |
| τ²-Bench Telecom (AA)not in index | – | – | – | 84.8%#96 | 92.1%#53 | – | – | Artificial Analysis | |
| MCP Atlas | – | – | – | – | 76.8%#16 | – | – | Scale AI SEAL | |
| HiL-Bench | – | – | – | – | – | – | 38.3%#8 | Scale AI SEAL | |
| APEX-Agents (AA)not in index | – | – | – | – | 33.0%#10 | – | – | Artificial Analysis | |
| Terminal-Bench 2.0 (Vals)not in index | – | – | – | – | – | 58.4%#10 | – | Vals AI | |
| DeepResearch Benchnot in index | – | 51.4%#8 | 53.2%#6 | 55.3%#1 | – | – | – | DeepResearch Bench | |
| METR Time Horizonsnot in index | – | – | – | – | – | – | 78.9%#2 | METR | |
| GBAEvalnot in index | – | – | – | – | – | – | 44.1%#11 | Epoch AI Benchmarking Hub | |
| Vending-Bench 2not in index | – | – | – | – | – | – | 8017.6#12 | Andon Labs |
Maths
| Benchmark | no reasoning | low | medium | high | max | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| FrontierMath Tiers 1–3 | – | – | – | – | 66.0%#28 | – | – | Epoch AI Benchmarking Hub | |
| FrontierMath Tier 4 | – | – | – | – | 26.8%#38 | – | – | Epoch AI Benchmarking Hub | |
| OTIS Mock AIME | – | – | – | – | 91.1%#62 | – | 94.4%#46 | Epoch AI Benchmarking Hub | |
| ProofBench | – | – | – | – | 50.0%#28 | – | – | Vals AI | |
| LiveBench Mathematicsnot in index | – | – | – | 89.3%#36 | – | – | – | LiveBench | |
| AIME (Vals) | – | – | – | – | – | 95.6%#8 | – | Vals AI | |
| AIME 2026 | – | – | – | 96.7%#6 | – | – | – | MathArena | |
| HMMT February 2026 | – | – | – | 96.2%#5 | – | – | – | MathArena | |
| MathArena Apex | – | – | – | 34.5%#8 | – | – | – | MathArena | |
| LMArena Maths | – | – | – | 1517#6 | – | – | 1507#10 | LMArena |
Knowledge
| Benchmark | no reasoning | low | medium | high | max | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| SimpleQA Verified | – | – | – | – | 47.0%#33 | – | – | Epoch AI Benchmarking Hub | |
| LiveBench Data Analysisnot in index | – | – | – | 69.9%#50 | – | – | – | LiveBench | |
| AA-Omniscience | – | – | – | 2.4#108 | 13.7#83 | – | – | Artificial Analysis | |
| MMLU-Pro | – | – | – | – | – | 89.1%#15 | – | Vals AI | |
| LegalBench | – | – | – | – | – | 85.3%#20 | – | Vals AI | |
| CorpFin | – | – | – | – | – | 67.0%#13 | – | Vals AI | |
| TaxEval | – | – | – | – | – | 76.0%#8 | – | Vals AI | |
| MedQA | – | – | – | – | – | 95.4%#12 | – | Vals AI | |
| PRBench Finance | 53.3%#4 | – | – | – | – | – | – | Scale AI SEAL | |
| PRBench Legal | 52.3%#5 | – | – | – | – | – | – | Scale AI SEAL | |
| MultiNRC | 48.3%#13 | – | – | – | 57.1%#8 | – | – | Scale AI SEAL | |
| CaseLaw v2not in index | – | – | – | – | – | 62.1%#20 | – | Vals AI | |
| MedCodenot in index | – | – | – | – | – | 49.1%#23 | 48.2%#28 | Vals AI | |
| MedScribenot in index | – | – | – | – | – | 86.1%#22 | 86.7%#19 | Vals AI | |
| DrugDiscoveryBenchnot in index | – | – | – | – | 27.7%#17 | – | – | Scale AI SEAL | |
| EBR-benchnot in index | – | – | – | – | 12.7%#17 | – | – | Epoch AI Benchmarking Hub | |
| AA-Omniscience: accuracynot in index | – | – | – | 45.8%#82 | 47.0%#76 | – | – | Artificial Analysis | |
| AA-Omniscience: non-hallucinationnot in index | – | – | – | 19.9%#264 | 37.2%#163 | – | – | Artificial Analysis |
Instruction following
| Benchmark | no reasoning | low | medium | high | max | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| LiveBench Languagenot in index | – | – | – | 83.3%#20 | – | – | – | LiveBench | |
| LiveBench Instruction Followingnot in index | – | – | – | 63.3%#47 | – | – | – | LiveBench | |
| IFBench | – | – | – | 44.6%#207 | 53.1%#156 | – | – | Artificial Analysis | |
| MultiChallenge | 56.0%#17 | – | – | – | 37.1%#28 | – | – | Scale AI SEAL | |
| LMArena Instruction Followingnot in index | – | – | – | 1514#3 | – | – | 1499#6 | LMArena |
Human preference
| Benchmark | no reasoning | low | medium | high | max | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| LMArena Text | – | – | – | 1505#2 | – | – | 1497#7 | LMArena | |
| EQ-Bench 4 | – | – | – | – | – | – | 1223#12 | EQ-Bench | |
| LMArena Creative Writingnot in index | – | – | – | 1501#4 | – | – | 1479#12 | LMArena | |
| LMArena Multi-turnnot in index | – | – | – | 1518#3 | – | – | 1512#7 | LMArena | |
| LMArena Longer Queriesnot in index | – | – | – | 1524#3 | – | – | 1517#5 | LMArena |
Multimodal
| Benchmark | no reasoning | low | medium | high | max | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| LMArena Vision | – | – | – | 1316#5 | – | – | 1313#9 | LMArena | |
| MMMU-Pro | – | – | – | 72.5%#126 | 75.4%#94 | – | – | Artificial Analysis | |
| VISTA | 45.5%#25 | – | – | – | 46.1%#22 | – | – | Scale AI SEAL | |
| MMMU-Pro (Vals)not in index | – | – | – | – | – | 83.9%#29 | – | Vals AI | |
| MortgageTaxnot in index | – | – | – | – | – | 68.5%#15 | – | Vals AI | |
| SAGEnot in index | – | – | – | – | – | 51.6%#12 | – | Vals AI | |
| VTBnot in index | – | – | – | – | – | 27.5%#4 | – | Scale AI SEAL |
Long context
| Benchmark | no reasoning | low | medium | high | max | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| AA-LCR | – | – | – | 67.0%#225 | 78.0%#110 | – | – | Artificial Analysis | |
| LMArena Documentnot in index | – | – | – | 1495#2 | – | – | 1494#4 | LMArena |
Composite
| Benchmark | no reasoning | low | medium | high | max | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| Epoch Capabilities Indexnot in index | – | – | – | – | – | – | 155.3#27 | Epoch AI Benchmarking Hub | |
| LiveBench | – | – | – | 74.5%#40 | – | – | – | LiveBench | |
| AA Intelligence Index | – | – | – | 26.4#137 | 31.9#100 | – | – | Artificial Analysis |
Safety & honesty
| Benchmark | no reasoning | low | medium | high | max | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|
| FORTRESSnot in index | 20.5%#37 | – | – | – | 13.0%#59 | – | – | Scale AI SEAL | |
| MASKnot in index | 96.3%#1 | – | – | – | 85.4%#18 | – | – | Scale AI SEAL |
Where it wins
Benchmarks where this configuration ranks in the top five of every configuration measured.
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 | 52 tok/s | 1.59 s | $5.50 | $27.50 | 1M | – |
| Google Vertex | 39 tok/s | 1.86 s | $5.00 | $25.00 | 1M | – |
| Anthropic | 39 tok/s | 1.67 s | $5.00 | $25.00 | 1M | – |
| Claude Platform on AWS | 38 tok/s | 1.63 s | $5.00 | $25.00 | 1M | – |
| Amazon Bedrock | 38 tok/s | 2.16 s | $5.00 | $25.00 | 1M | – |
| Azure | 34 tok/s | 2.60 s | $5.00 | $25.00 | 1M | – |
Price history
Listed price per 1M tokens over time, as recorded by OpenRouter for the provider with the longest history.
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 $0.500 per 1M. Reasoning tokens are not modelled.
| Workload | Tokens in / out | Cost | With caching | Time |
|---|---|---|---|---|
| Chat reply | 400 / 300 | $0.0095 | $0.0081 | 9.4 s |
| Summarise a 30-page report | 12,000 / 600 | $0.075 | $0.035 | 17.3 s |
| Code edit | 6,000 / 1,500 | $0.068 | $0.047 | 41.0 s |
| Agentic coding session | 60,000 / 4,000 | $0.400 | $0.198 | 1.8 min |
| Structured extraction | 2,000 / 200 | $0.015 | $0.0083 | 6.8 s |
See also
Data as of 5 Oct 2026. Compare these configurations.
Cite as: BenchLeader, “Claude Opus 4.6: benchmarks, pricing, speed and rank”, https://www.benchleader.com/models/claude-opus-4-6, data as of 5 Oct 2026.