Capability
82.7/100
field median 58
#2 of 224 ranked models
Claude Opus 5
Released Jul 24, 2026 — see all recent releases
Decision reading
Claude Opus 5 scores 82.7 out of 100 and ranks #2 of 224. This profile shows 63 source-displayable benchmark rows; its strongest eligible category is Agentic at #1. API pricing is $5 input and $25 output per million tokens, with cached input at $0.5.
Data as of August 22, 2026 · How the score is built
Strongest published evidence
Agentic ranks #1. Particularly useful for coding agents, browser research, and computer-use workflows.
Validate before choosing
63 published rows leave some tracked benchmark slots empty. Coding is its lowest eligible category at #5.
Decision snapshot
Each value carries a field reference instead of floating alone. Markers compare this model with the current ranked and priced catalog; they are not absolute quality thresholds.
Price
$5input / $25 output
input median $1
cached $0.50 · batch + cache $0.25 · blended $15
Speed
62tok/s
field median 91.5 tok/s
First token 52.65 s
Context
1Mtokens
field median 200,000
Maximum output length is tracked separately
Capability shape
Each axis shows percentile within that category’s eligible cohort. The comparison outline is the median of the six nearest public-score peers; a collapsed vertex means the category is not rank-eligible.
Claude Opus 5 category percentile values
- Agentic100th percentile
- Coding97th percentile
- ReasoningNot eligible
- Knowledge100th percentile
- MathNot eligible
- MultilingualNot eligible
- Multimodal100th percentile
- Instruction followingNot eligible
The dashed outline is median of 6 nearest peers.
Eligible category ranks
- Agentic#1/137
- Coding#5/141
- ReasoningNot ranked
- Knowledge#1/60
- MathNot ranked
- MultilingualNot ranked
- Multimodal#1/34
- Inst. FollowingNot ranked
Top decileTop quartileMid-fieldNot eligible
What it costs to get this score
Published API price against the public score. The x-axis uses a log scale; the dashed path marks models that are not beaten by a cheaper, higher-scoring option. Price uses average of published input and output rates.
The chart opens on the current model. Scroll horizontally to inspect the full price axis.
Current modelClaude Opus 5 · 82.7 score · $15 blended per million tokens
Horizontal: blended price per million tokens, log scale · Vertical: public score
How much of this is verified
Coverage is split by category so a strong number never hides a thin evidence base. Verified means the row is tied to a published source; provisional rows remain visible but separate.
- Agentic18/18 verified
- Coding11/11 verified
- Reasoning3/3 verified
- Knowledge15/15 verified
- Math5/5 verified
- Multilingual3/3 verified
- Multimodal8/8 verified
- Inst. FollowingNot measured
Verified sourceProvisionalNot measured
Spec sheet
Each documented value carries its source. Missing fields stay visible as not sourced or not published, rather than disappearing from the page.
- API model ID
- claude-opus-5Anthropic model overview
- Context window
- 1MAnthropic model overview
- Maximum output
- Not sourced yet
- Knowledge cutoff
- Not sourced yet
- Input modalities
- text, imageAnthropic model overview
- Output modalities
- textAnthropic model overview
- Parameters
- Not disclosed by the provider
- Availability
- Claude APIAnthropic model overview
- Cloud regions
- Not tracked yet
- Lifecycle
- activeAnthropic model overview
- API capabilities
- Tool calling, structured outputs, and batch support are not tracked yet
- Prompt caching
- Published at $0.50 per million cached input tokensClaude API pricing
- Self-host
- Weights are not published
- Rate limits
- Not tracked yet
Category score record
Scores and ranks appear only where published evidence can be displayed. The table keeps the score, weight, cohort, and evidence state together.
| Category | Score | Rank | Percentile | Weight | Benchmarks | Evidence |
|---|---|---|---|---|---|---|
| AgenticRank #1 of 137Percentile 100thWeight 22%18 benchmarksVerified | 79.7 | #1 of 137 | 100th | 22% | 18 benchmarks | Verified |
| CodingRank #5 of 141Percentile 97thWeight 20%11 benchmarksVerified | 78.3 | #5 of 141 | 97th | 20% | 11 benchmarks | Verified |
| ReasoningRank Not rankedWeight 17%3 benchmarksVerified | 90.8 | Not ranked | Not available | 17% | 3 benchmarks | Verified |
| KnowledgeRank #1 of 60Percentile 100thWeight 12%15 benchmarksVerified | 93.5 | #1 of 60 | 100th | 12% | 15 benchmarks | Verified |
| MathWeight 5%5 benchmarksVerified | Score pending | Not ranked | Not available | 5% | 5 benchmarks | Verified |
| MultilingualWeight 7%3 benchmarksVerified | Score pending | Not ranked | Not available | 7% | 3 benchmarks | Verified |
| MultimodalRank #1 of 34Percentile 100thWeight 12%8 benchmarksVerified | 88.5 | #1 of 34 | 100th | 12% | 8 benchmarks | Verified |
| Inst. FollowingWeight 5%0 benchmarksNot measured | Not measured | Not ranked | Not available | 5% | 0 benchmarks | Not measured |
Benchmark ledger
Coding opens by default. The marker compares each value with the best source-verified result in the catalog; provisional leaders do not set the reference. Expand the remaining categories for every published row.
Coding11 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| SWE-bench VerifiedSoftware Engineering Benchmark Verified | Score96% | Versus best verified row Best verified: Claude Opus 5 · 96% | GapBest verified | WeightWeighted 16% | Provider exact |
| SWE-bench Pro | Score79.2% | Versus best verified row Best verified: Claude Mythos 5 · 80.3% | Gap1.1 behind | WeightWeighted 10% | Provider exact |
| SWE Multilingual | Score89.5% | Versus best verified row Best verified: Claude Opus 5 · 89.5% | GapBest verified | WeightDisplay only | Provider exact |
| SWE MultimodalSWE-bench Multimodal | Score59.4% | Versus best verified row Best verified: Claude Opus 5 · 59.4% | GapBest verified | WeightDisplay only | Provider exact |
| deepSwe | Score68.8% | Versus best verified row Best verified: GPT-5.6 Sol · 72.7% | Gap3.9 behind | WeightDisplay only | Provider exact |
| FrontierCode 1.1 Main | Score53.4% | Versus best verified row Best verified: Claude Fable 5 · 53.5% | Gap0.1 behind | WeightDisplay only | Provider exact |
| FrontierCode 1.1 Extended | Score63.6% | Versus best verified row Best verified: Claude Opus 5 · 63.6% | GapBest verified | WeightDisplay only | Provider exact |
| ProgramBench (episode 1)ProgramBench hidden-test pass rate after episode 1 | Score83.0% | Versus best verified row Best verified: Claude Opus 5 · 83.0% | GapBest verified | WeightDisplay only | Provider exact |
| ProgramBenchProgramBench: Can Language Models Rebuild Programs From Scratch? | Score93.0% | Versus best verified row Best verified: Claude Opus 5 · 93.0% | GapBest verified | WeightDisplay only | Provider exact |
| cursorBench32 | Score70.0% | Versus best verified row Best verified: Grok 4.6 · 70.8% | Gap0.8 behind | WeightDisplay only | Benchmark exact |
| VulcanBench v3 | Score87.0% | Versus best verified row Best verified: Grok 4.5 · 91.3% | Gap4.3 behind | WeightDisplay only | Benchmark exact |
Agentic18 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| BrowseComp | Score90.8% | Versus best verified row Best verified: GPT-5.6 Sol · 92.2% | Gap1.4 behind | WeightWeighted 28% | Provider exact |
| Terminal-Bench 3.0 | Score42.7% | Versus best verified row Best verified: Claude Opus 5 · 42.7% | GapBest verified | WeightDisplay only | Benchmark exact |
| HLE w/ toolsHumanity's Last Exam with tools | Score64.7% | Versus best verified row Best verified: Claude Opus 5 · 64.7% | GapBest verified | WeightDisplay only | Provider exact |
| DeepSearchQA | Score95.0% | Versus best verified row Best verified: Claude Opus 5 · 95.0% | GapBest verified | WeightDisplay only | Provider exact |
| DRACOData Research and Analysis with Complex Operations | Score88.6% | Versus best verified row Best verified: Claude Opus 5 · 88.6% | GapBest verified | WeightDisplay only | Provider exact |
| BrowseComp (10-agent, prerelease)Multi-Agent BrowseComp — 10-agent team prerelease configuration | Score93.6% | Versus best verified row Best verified: Claude Opus 5 · 93.6% | GapBest verified | WeightDisplay only | Provider exact |
| OSWorld 2.0 | Score70.6% | Versus best verified row Best verified: Claude Opus 5 · 70.6% | GapBest verified | WeightDisplay only | Provider exact |
| MCP Atlas | Score85.8% | Versus best verified row Best verified: Muse Spark 1.1 · 88.1% | Gap2.3 behind | WeightDisplay only | Provider exact |
| MCP-Atlas claim coverageMCP-Atlas mean claim coverage | Score89.1% | Versus best verified row Best verified: Claude Opus 5 · 89.1% | GapBest verified | WeightDisplay only | Provider exact |
| LAB all-pass (Anthropic harness)Legal Agent Benchmark all-pass rate — Anthropic harness | Score23.58% | Versus best verified row Best verified: Claude Opus 5 · 23.58% | GapBest verified | WeightDisplay only | Provider exact |
| LAB criterion-pass (Anthropic harness)Legal Agent Benchmark mean criterion-pass rate — Anthropic harness | Score93.74% | Versus best verified row Best verified: Claude Opus 5 · 93.74% | GapBest verified | WeightDisplay only | Provider exact |
| LAB all-pass (Harvey held-out)Legal Agent Benchmark all-pass rate — Harvey held-out set | Score11.7% | Versus best verified row Best verified: Claude Opus 5 · 11.7% | GapBest verified | WeightDisplay only | Provider exact |
| LAB criterion-pass (Harvey held-out)Legal Agent Benchmark mean criterion-pass rate — Harvey held-out set | Score94.1% | Versus best verified row Best verified: Claude Opus 5 · 94.1% | GapBest verified | WeightDisplay only | Provider exact |
| Toolathlon-Verified | Score80.6% | Versus best verified row Best verified: Claude Opus 5 · 80.6% | GapBest verified | WeightDisplay only | Provider exact |
| Toolathlon Verified Pass@3 | Score87.0% | Versus best verified row Best verified: Claude Opus 5 · 87.0% | GapBest verified | WeightDisplay only | Provider exact |
| Toolathlon Verified Pass³Toolathlon Verified Pass cubed | Score73.1% | Versus best verified row Best verified: Claude Opus 5 · 73.1% | GapBest verified | WeightDisplay only | Provider exact |
| Toolathlon Verified avg. turnsToolathlon Verified average assistant turns | Score23.5 turns | Versus best verified row Best verified: Claude Opus 5 · 23.5 turns | GapBest verified | WeightDisplay only | Provider exact |
| AutomationBench | Score26.0% | Versus best verified row Best verified: GLM-5.3 · 48.2% | Gap22.2 behind | WeightDisplay only | Provider exact |
Reasoning3 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| ARC-AGI-2Abstraction and Reasoning Corpus for AGI v2 | Score90.4% | Versus best verified row Best verified: GPT-5.6 Sol · 92.5% | Gap2.1 behind | WeightWeighted 31% | Provider exact |
| ARC-AGI-1ARC-AGI-1 Semi-Private Evaluation | Score97.50% | Versus best verified row Best verified: Claude Opus 5 · 97.50% | GapBest verified | WeightDisplay only | Provider exact |
| ARC-AGI-3Abstraction and Reasoning Corpus for AGI v3 | Score30.2% | Versus best verified row Best verified: Claude Opus 5 · 30.2% | GapBest verified | WeightDisplay only | Provider exact |
Knowledge15 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| HLEHumanity's Last Exam | Score64.7% | Versus best verified row Best verified: Claude Opus 5 · 64.7% | GapBest verified | WeightWeighted 45% | Provider exact |
| HLE w/o toolsHumanity's Last Exam without tools | Score56.3% | Versus best verified row Best verified: Claude Mythos 5 · 59% | Gap2.7 behind | WeightDisplay only | Provider exact |
| HealthBench (raw)HealthBench raw score | Score67.1% | Versus best verified row Best verified: Claude Opus 5 · 67.1% | GapBest verified | WeightDisplay only | Provider exact |
| HealthBench (length-adjusted)HealthBench length-adjusted score | Score57.8% | Versus best verified row Best verified: Claude Opus 5 · 57.8% | GapBest verified | WeightDisplay only | Provider exact |
| HealthBench Professional | Score59.8% | Versus best verified row Best verified: GPT-5.6 Sol · 60.5% | Gap0.7 behind | WeightDisplay only | Provider exact |
| HealthBench Professional (raw)HealthBench Professional raw score | Score73.4% | Versus best verified row Best verified: Claude Opus 5 · 73.4% | GapBest verified | WeightDisplay only | Provider exact |
| BioMysteryBench (human-solvable)BioMysteryBench Human Solvable | Score90.1% | Versus best verified row Best verified: Claude Opus 5 · 90.1% | GapBest verified | WeightDisplay only | Provider exact |
| BioMysteryBench (human-difficult)BioMysteryBench Human Difficult | Score49.4% | Versus best verified row Best verified: Claude Opus 5 · 49.4% | GapBest verified | WeightDisplay only | Provider exact |
| SpatialBench VerifiedLatchBio SpatialBench Verified | Score72.5% | Versus best verified row Best verified: Claude Opus 5 · 72.5% | GapBest verified | WeightDisplay only | Provider exact |
| SingleCellBenchLatchBio SingleCellBench | Score60.6% | Versus best verified row Best verified: Claude Opus 5 · 60.6% | GapBest verified | WeightDisplay only | Provider exact |
| ProteinGym Hard | Score47.7% | Versus best verified row Best verified: Claude Opus 5 · 47.7% | GapBest verified | WeightDisplay only | Provider exact |
| Protein DesignAnthropic Protein Design evaluation | Score42.5% | Versus best verified row Best verified: Claude Opus 5 · 42.5% | GapBest verified | WeightDisplay only | Provider exact |
| Organic chemistry V2Anthropic Organic Chemistry V2 evaluation | Score61.6% | Versus best verified row Best verified: Claude Opus 5 · 61.6% | GapBest verified | WeightDisplay only | Provider exact |
| Protocols (troubleshooting)Molecular Biology Protocols Troubleshooting | Score61.1% | Versus best verified row Best verified: Claude Opus 5 · 61.1% | GapBest verified | WeightDisplay only | Provider exact |
| Protocols (understanding)Benchling Molecular Biology Protocols Understanding | Score78.4% | Versus best verified row Best verified: Claude Opus 5 · 78.4% | GapBest verified | WeightDisplay only | Provider exact |
Math5 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| IMO 2026International Mathematical Olympiad 2026 | Score42/42 | Versus best verified row Best verified: Claude Opus 5 · 42/42 | GapBest verified | WeightDisplay only | Provider exact |
| RiemannBench (no tools)RiemannBench without tools | Score60.0% | Versus best verified row Best verified: Claude Opus 5 · 60.0% | GapBest verified | WeightDisplay only | Provider exact |
| RiemannBench (tools)RiemannBench with tools | Score79.0% | Versus best verified row Best verified: Claude Opus 5 · 79.0% | GapBest verified | WeightDisplay only | Provider exact |
| ArXivMath Jun. 2026 (no tools)ArXivMath June 2026 without tools | Score90.8% | Versus best verified row Best verified: Claude Opus 5 · 90.8% | GapBest verified | WeightDisplay only | Provider exact |
| ArXivMath Jun. 2026 (tools)ArXivMath June 2026 with tools | Score91.3% | Versus best verified row Best verified: Claude Opus 5 · 91.3% | GapBest verified | WeightDisplay only | Provider exact |
Multilingual3 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| GMMLUGlobal MMLU | Score92.5% | Versus best verified row Best verified: Claude Opus 5 · 92.5% | GapBest verified | WeightDisplay only | Provider exact |
| MILUMulti-task Indic Language Understanding Benchmark | Score92.1% | Versus best verified row Best verified: Claude Opus 5 · 92.1% | GapBest verified | WeightDisplay only | Provider exact |
| INCLUDE | Score89.8% | Versus best verified row Best verified: Claude Opus 5 · 89.8% | GapBest verified | WeightDisplay only | Provider exact |
Multimodal8 rows
| Benchmark | Score | Versus best verified row | Gap | Weight | Evidence |
|---|---|---|---|---|---|
| OfficeQA Pro | Score66.9% | Versus best verified row Best verified: Claude Opus 5 · 66.9% | GapBest verified | WeightWeighted 30% | Provider exact |
| Chartography (no tools)Chartography without tools | Score29.6% | Versus best verified row Best verified: Claude Opus 5 · 29.6% | GapBest verified | WeightDisplay only | Provider exact |
| Chartography (tools)Chartography with image and code tools | Score83.0% | Versus best verified row Best verified: Claude Opus 5 · 83.0% | GapBest verified | WeightDisplay only | Provider exact |
| BenchCAD Vision2Code (no tools)BenchCAD Vision2Code voxel IoU without tools | Score0.366 | Versus best verified row Best verified: Claude Opus 5 · 0.366 | GapBest verified | WeightDisplay only | Provider exact |
| BenchCAD Vision2Code (tools)BenchCAD Vision2Code voxel IoU with tools | Score0.821 | Versus best verified row Best verified: Claude Opus 5 · 0.821 | GapBest verified | WeightDisplay only | Provider exact |
| GDP.pdf (no tools)GDP.pdf mean criteria pass rate without tools | Score83.4% | Versus best verified row Best verified: Claude Opus 5 · 83.4% | GapBest verified | WeightDisplay only | Provider exact |
| GDP.pdf (tools)GDP.pdf mean criteria pass rate with tools | Score85.5% | Versus best verified row Best verified: Claude Opus 5 · 85.5% | GapBest verified | WeightDisplay only | Provider exact |
| OfficeQA | Score78.1% | Versus best verified row Best verified: Claude Opus 5 · 78.1% | GapBest verified | WeightDisplay only | Provider exact |
Lineage
The sequence follows explicit supersedes links. Scores and prices remain blank when the corresponding public row or first-party rate is unavailable.
February 2026
Claude Opus 4.6Score 67.8 · $5 / $25
Apr 16, 2026
Claude Opus 4.7 (Adaptive)Score 72.3 · $5 / $25
May 28, 2026
Claude Opus 4.8Score 76.2 · $5 / $25
Jul 24, 2026 · you are here
Claude Opus 5Score 82.7 · $5 / $25
Base entry
How to read this profile
The visual layer above carries the decisions. These notes preserve the model, ranking, coverage, and family context behind the numbers.
Claude Opus 5 ranks #2 of 224 on the public leaderboard with a score of 82.72/100. Its source-verified position is #2 of 104.
Claude Opus 5 is a proprietary model with a 1M context window. It uses an explicit reasoning mode, which can improve complex problem solving while adding latency and token use.
Available on all Claude platforms and through the Claude API as claude-opus-5. It is the default model on Claude Max and the strongest model offered on Claude Pro. Anthropic lists Opus 5 at $5 input / $25 output per million tokens, with Fast mode running around 2.5× the default speed at twice the base price.
Official exact-value snapshot from Anthropic's July 24, 2026 Claude Opus 5 launch post and system card, supplemented by Cursor's live CursorBench 3.2 Max row. BenchLM maps the highest published effort setting into the canonical row and uses the verified high-effort ARC-AGI-3 result because no max-effort score was available at launch.
Its explicit predecessor is Claude Opus 4.8. 63 of 402 tracked benchmark slots currently have displayable evidence. Missing categories stay blank.
Its strongest eligible category is Agentic at #1, while its lowest eligible position is Coding at #5. particularly useful for coding agents, browser research, and computer-use workflows.
Radar
Full release historyClaude Opus 5 release history
Claude Opus 5 released
Anthropic · Model release
Source confirmed
Frequently asked questions
How does Claude Opus 5 perform overall in AI benchmarks?
Claude Opus 5 ranks #2 out of 224 models on the public BenchAlign leaderboard, with a score of 82.72/100. Its evidence status is Supported, and this profile shows 63 source-displayable benchmark rows. The label describes evidence depth, not a provider quality claim; inspect category rows before choosing a workload.
Is Claude Opus 5 good for knowledge and understanding?
Claude Opus 5 ranks #1 out of 60 eligible models for knowledge and understanding, with a public category score of 93.5/100. That places it in the current top ten for this category. Check the underlying rows before treating the aggregate as a workload guarantee.
Is Claude Opus 5 good for coding and programming?
Claude Opus 5 ranks #5 out of 141 eligible models for coding and programming, with a public category score of 78.3/100. That places it in the current top ten for this category. Check the underlying rows before treating the aggregate as a workload guarantee.
Is Claude Opus 5 good for mathematics?
Claude Opus 5 has source-displayable benchmark coverage for mathematics, but the public category table does not assign it a rank there. The individual rows remain available for inspection. A missing category position means the evidence threshold was not met; it does not convert the model's unmeasured work into a zero.
Is Claude Opus 5 good for reasoning and logic?
Claude Opus 5 has source-displayable benchmark coverage for reasoning and logic, but the public category table does not assign it a rank there. The individual rows remain available for inspection. A missing category position means the evidence threshold was not met; it does not convert the model's unmeasured work into a zero.
Is Claude Opus 5 good for agentic tool use and computer tasks?
Claude Opus 5 ranks #1 out of 137 eligible models for agentic tool use and computer tasks, with a public category score of 79.7/100. That places it in the current top ten for this category. Check the underlying rows before treating the aggregate as a workload guarantee.
Is Claude Opus 5 good for multimodal and grounded tasks?
Claude Opus 5 ranks #1 out of 34 eligible models for multimodal and grounded tasks, with a public category score of 88.5/100. That places it in the current top ten for this category. Check the underlying rows before treating the aggregate as a workload guarantee.
Is Claude Opus 5 good for multilingual tasks?
Claude Opus 5 has source-displayable benchmark coverage for multilingual tasks, but the public category table does not assign it a rank there. The individual rows remain available for inspection. A missing category position means the evidence threshold was not met; it does not convert the model's unmeasured work into a zero.
Does Claude Opus 5 have full benchmark coverage on BenchLM?
No. Claude Opus 5 currently has 82 source-displayable rows across 402 tracked benchmark slots. The profile exposes published, non-generated evidence and leaves missing categories blank until an exact evaluation is available. Coverage describes how much was measured; it is not a penalty added to an individual benchmark result.
What is the context window size of Claude Opus 5?
Claude Opus 5 has a documented context window of 1M. That figure is the maximum combined prompt and retained-conversation space reported for this exact model; it is not the maximum output length. The profile keeps output limits separate because providers often publish those limits independently.
Last updated August 22, 2026. Runtime fields remain blank until a sourced snapshot exists.
Watch Claude Opus 5 in the weekly brief
Get one weekly email when material rank, price, availability, or benchmark evidence changes are worth revisiting.
Read a sample issueJoin 2,000+ readers.