AnthropicReasoning modelClaude Opus 4.8
Claude Opus 4.8 is an Anthropic proprietary reasoning model, released 28 May 2026. Its best configuration (max reasoning effort) ranks #82 of 756 on the BenchLeader Index at 62.4 ±4.8, in the upper half. The ± is the point: 173 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 (69) and lowest in agents & tools (57). At $10.00 per million tokens blended it is among the most expensive ranked models. Output speed of 61 tokens per second puts it slower than most, with a first answer in 63.3 s. It has been measured at 8 reasoning-effort settings; this summary describes the best-scoring one, and the tabs above switch between them. Last measured 25 Jun 2026.
- Blended price
- $10.00/M
- $5.00 in · $25.00 out
- Output speed
- 61 tok/s
- measured by Artificial Analysis
- First answer
- 63 s
- first token 1.28 s
- Context
- 1M
- Full answer
- 71 s
- median, reasoning included
- Cost per run
- $4.08
- one full Intelligence Index run
- Released
- 28 May 2026
- Overall index62
- Reasoning69
- Coding60
- Agents & tools57
- Maths68
- Knowledge67
- Instruction following62
- Long context64
- Composite65
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 4.8maxthis page | 28 May 2026 | 62.4 | #82 |
| Claude Opus 4.7 | 16 Apr 2026 | 63.8 | #59 |
| Claude Opus 4.6 | 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 | – | – | 57 tok/s | 1.28 s | $0.0095 | Coding 64 · Maths 61 · Reasoning 62 |
| low | – | – | 57 tok/s | 1.28 s | $0.0095 | Maths 67 · Reasoning 62 |
| medium | – | – | 57 tok/s | 1.28 s | $0.0095 | Coding 68 · Reasoning 64 |
| high | 60.9 | #100 | 57 tok/s | 1.28 s | $0.0095 | Agents & tools 61 · Coding 62 · Human preference 68 · Maths 68 · Multimodal 64 · Reasoning 61 |
| xhigh | – | – | 57 tok/s | 1.28 s | $0.0095 | Coding 72 · Reasoning 30 |
| maxbest | 62.4 | #82 | 61 tok/s | 63 s | $0.0095 | Agents & tools 57 · Coding 60 · Composite 65 · Instruction following 62 · Knowledge 67 · Long context 64 · Maths 68 · Reasoning 69 |
| thinking | – | – | 57 tok/s | 1.28 s | $0.0095 | Reasoning 80 |
| not stated | 60.9 | #98 | 57 tok/s | 1.28 s | $0.0095 | Agents & tools 55 · Coding 62 · Human preference 66 · Knowledge 62 · Maths 66 · Multimodal 64 · Reasoning 65 |
How its index has moved
8 Sept 2026 to 5 Oct 2026- max
- high
- 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
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
| Benchmark | no reasoning | low | medium | high | xhigh | max | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|---|
| SciCode | – | – | – | – | – | 53.5%#69 | – | – | SciCode | |
| WeirdML | 70.5%#32 | – | 76.0%#27 | – | 82.9%#17 | – | – | – | WeirdML | |
| FrontierCode | – | – | – | – | – | – | – | 46.5%#13 | Cognition | |
| GSO-Bench | – | – | – | – | – | – | – | 47.1%#5 | GSO-Bench | |
| LMArena Coding | – | – | – | 1533#16 | – | – | – | 1530#21 | LMArena | |
| LMArena WebDev | – | – | – | 1556#34 | – | – | – | 1534#42 | LMArena | |
| LiveBench Codingnot in index | – | – | – | – | – | 81.8%#12 | – | – | LiveBench | |
| SciCode (AA)not in index | – | – | – | – | – | 54.4%#57 | – | – | Artificial Analysis | |
| LiveCodeBench | – | – | – | – | – | – | – | 87.8%#11 | Vals AI | |
| SWE-bench (Vals)not in index | – | – | – | – | – | – | – | 88.6%#12 | Vals AI | |
| Code Migrationnot in index | – | – | – | – | – | – | – | 47.3%#13 | Vals AI | |
| ProgramBenchnot in index | – | – | – | – | – | – | – | 1.0%#11 | Vals AI | |
| Vibe Code Bench v1.1not in index | – | – | – | – | – | – | – | 82.7%#16 | Vals AI | |
| SWE Atlas: Codebase QnAnot in index | – | – | – | – | 57.3%#4 | – | – | – | Scale AI SEAL | |
| SWE Atlas: Refactoringnot in index | – | – | – | – | – | – | – | 46.7%#5 | Scale AI SEAL | |
| SWE Atlas: Test Writingnot in index | – | – | – | – | 49.6%#6 | – | – | – | Scale AI SEAL | |
| Surface Evolver Benchnot in index | 68.1%#11 | – | – | 87.5%#6 | – | – | – | – | Epoch AI Benchmarking Hub | |
| DeepSWEnot in index | – | 40.8%#56 | 48.7%#49 | 51.8%#46 | 54.4%#39 | 59.0%#34 | – | – | Epoch AI Benchmarking Hub | |
| ALE-Benchnot in index | 1411.8#21 | – | – | 1563.8#16 | – | – | – | – | Epoch AI Benchmarking Hub | |
| Terminal-Bench 4.0 (AA)not in index | – | – | – | – | – | 21.7%#50 | – | – | Artificial Analysis | |
| Terminal-Bench 2.1 (AA)not in index | – | – | – | – | – | 84.6%#28 | – | – | Artificial Analysis |
Agents & tools
Maths
| Benchmark | no reasoning | low | medium | high | xhigh | max | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|---|
| FrontierMath Tiers 1–3 | – | – | – | – | – | 80.0%#15 | – | – | Epoch AI Benchmarking Hub | |
| FrontierMath Tier 4 | – | – | – | – | – | 56.1%#22 | – | – | Epoch AI Benchmarking Hub | |
| OTIS Mock AIME | 84.4%#90 | 97.8%#24 | – | – | – | 98.3%#20 | – | – | Epoch AI Benchmarking Hub | |
| ProofBench | – | – | – | – | – | 69.0%#15 | – | – | Vals AI | |
| LiveBench Mathematicsnot in index | – | – | – | – | – | 94.3%#18 | – | – | LiveBench | |
| AIME 2026 | – | – | – | – | – | 100.0%#1 | – | – | MathArena | |
| HMMT February 2026 | – | – | – | – | – | 95.5%#6 | – | – | MathArena | |
| MathArena Apex | – | – | – | – | – | 81.3%#1 | – | – | MathArena | |
| LMArena Maths | – | – | – | 1494#21 | – | – | – | 1477#38 | LMArena |
Knowledge
| Benchmark | no reasoning | low | medium | high | xhigh | max | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|---|
| SimpleQA Verified | – | – | – | – | – | 53.0%#20 | – | – | Epoch AI Benchmarking Hub | |
| LiveBench Data Analysisnot in index | – | – | – | – | – | 66.0%#54 | – | – | LiveBench | |
| AA-Omniscience | – | – | – | – | – | 28.8#35 | – | – | Artificial Analysis | |
| MMLU-Pro | – | – | – | – | – | – | – | 89.6%#9 | Vals AI | |
| LegalBench | – | – | – | – | – | – | – | 83.6%#50 | Vals AI | |
| CorpFin | – | – | – | – | – | – | – | 66.7%#17 | Vals AI | |
| TaxEval | – | – | – | – | – | – | – | 75.6%#14 | Vals AI | |
| Excel Modeling Benchmarknot in index | – | – | – | – | – | – | – | 69.4%#13 | Vals AI | |
| MedCodenot in index | – | – | – | – | – | – | – | 53.2%#10 | Vals AI | |
| MedScribenot in index | – | – | – | – | – | – | – | 85.8%#24 | Vals AI | |
| Public Benefits Benchnot in index | – | – | – | – | – | – | – | 68.1%#10 | Vals AI | |
| DrugDiscoveryBenchnot in index | – | – | – | – | – | 46.8%#8 | – | – | Scale AI SEAL | |
| EBR-benchnot in index | – | – | – | – | – | 28.6%#11 | – | – | Epoch AI Benchmarking Hub | |
| GDP.pdfnot in index | – | – | – | – | – | 24.0%#15 | – | – | Epoch AI Benchmarking Hub | |
| AA-Omniscience: accuracynot in index | – | – | – | – | – | 48.8%#67 | – | – | Artificial Analysis | |
| AA-Omniscience: non-hallucinationnot in index | – | – | – | – | – | 60.8%#67 | – | – | Artificial Analysis |
Instruction following
| Benchmark | no reasoning | low | medium | high | xhigh | max | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|---|
| LiveBench Languagenot in index | – | – | – | – | – | 79.7%#34 | – | – | LiveBench | |
| LiveBench Instruction Followingnot in index | – | – | – | – | – | 72.0%#19 | – | – | LiveBench | |
| IFBench | – | – | – | – | – | 62.2%#116 | – | – | Artificial Analysis | |
| LMArena Instruction Followingnot in index | – | – | – | 1491#11 | – | – | – | 1478#22 | LMArena |
Human preference
| Benchmark | no reasoning | low | medium | high | xhigh | max | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|---|
| LMArena Text | – | – | – | 1482#24 | – | – | – | 1475#37 | LMArena | |
| EQ-Bench 4 | – | – | – | – | – | – | – | 1281#6 | EQ-Bench | |
| LMArena Creative Writingnot in index | – | – | – | 1470#16 | – | – | – | 1463#23 | LMArena | |
| LMArena Multi-turnnot in index | – | – | – | 1497#11 | – | – | – | 1494#16 | LMArena | |
| LMArena Longer Queriesnot in index | – | – | – | 1503#12 | – | – | – | 1500#15 | LMArena |
Multimodal
| Benchmark | no reasoning | low | medium | high | xhigh | max | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|---|
| LMArena Vision | – | – | – | 1295#22 | – | – | – | 1289#27 | LMArena | |
| MMMU-Pro (Vals)not in index | – | – | – | – | – | – | – | 86.6%#18 | Vals AI | |
| MortgageTaxnot in index | – | – | – | – | – | – | – | 69.9%#5 | Vals AI | |
| SAGEnot in index | – | – | – | – | – | – | – | 54.8%#3 | Vals AI | |
| Blueprint-Bench 2not in index | – | – | – | – | – | – | – | 14.5%#24 | Epoch AI Benchmarking Hub |
Long context
| Benchmark | no reasoning | low | medium | high | xhigh | max | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|---|
| AA-LCR | – | – | – | – | – | 77.7%#116 | – | – | Artificial Analysis | |
| LMArena Documentnot in index | – | – | – | 1486#9 | – | – | – | 1475#16 | LMArena |
Composite
| Benchmark | no reasoning | low | medium | high | xhigh | max | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|---|
| Epoch Capabilities Indexnot in index | – | – | – | – | – | – | – | 158.3#12 | Epoch AI Benchmarking Hub | |
| LiveBench | – | – | – | – | – | 76.2%#29 | – | – | LiveBench | |
| AA Intelligence Index | – | – | – | – | – | 41.8#51 | – | – | Artificial Analysis | |
| Vals Indexnot in index | – | – | – | – | – | – | – | 55.1#13 | Vals AI | |
| Vals Multimodal Indexnot in index | – | – | – | – | – | – | – | 70.9%#5 | Vals AI |
Safety & honesty
| Benchmark | no reasoning | low | medium | high | xhigh | max | thinking | not stated | Source | Trend |
|---|---|---|---|---|---|---|---|---|---|---|
| FORTRESSnot in index | – | – | – | – | – | 18.2%#41 | – | – | 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 |
|---|---|---|---|---|---|---|
| Anthropic | 169 tok/s | 0.92 s | $10.00 | $50.00 | 1M | – |
| Google Vertex | 81 tok/s | 2.71 s | $5.00 | $25.00 | 1M | – |
| Amazon Bedrock | 72 tok/s | 6.20 s | $5.50 | $27.50 | 1M | – |
| Amazon Bedrock | 66 tok/s | 3.70 s | $5.00 | $25.00 | 1M | – |
| Claude Platform on AWS | 62 tok/s | 1.44 s | $5.00 | $25.00 | 1M | – |
| Anthropic | 51 tok/s | 1.65 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 | 1.1 min |
| Summarise a 30-page report | 12,000 / 600 | $0.075 | $0.035 | 1.2 min |
| Code edit | 6,000 / 1,500 | $0.068 | $0.047 | 1.5 min |
| Agentic coding session | 60,000 / 4,000 | $0.400 | $0.198 | 2.1 min |
| Structured extraction | 2,000 / 200 | $0.015 | $0.0083 | 1.1 min |
See also
Data as of 5 Oct 2026. Compare these configurations.
Cite as: BenchLeader, “Claude Opus 4.8: benchmarks, pricing, speed and rank”, https://www.benchleader.com/models/claude-opus-4-8, data as of 5 Oct 2026.