Claude Opus 4
Anthropic · released May 22, 2025
Claude Opus 4 is benchmarked as the world’s best coding model, at time of release, bringing sustained performance on complex, long-running tasks and agent workflows. It sets new benchmarks in...
Specification
- Context window
- 200K
- Max output
- 32K
- Knowledge cutoff
- Jan 31, 2025
- Parameters
- Undisclosed
- Licence
- Proprietary
- Serving providers
- 1
- Moderated
- No
- Uptime
- 100.0%
Intelligence
26.0
69th percentile
Coding
—
Coding Index
Agentic
—
Agentic Index
Output speed
22 t/s
Median across providers
Latency
2.22s
Time to first token
Cost per task
$6.75
Estimated
Benchmarks
Where the score comes from
The Intelligence Index is a composite. These are the underlying evaluations this model was actually measured on.
Evaluation scores
Percentage correct · higher is better
- MMLU-Pro86.0%
- GPQA Diamond70.1%
- LiveCodeBench54.2%
- IFBench43.3%
- SciCode40.9%
- AA-LCR (long context)40.0%
- AIME 202536.3%
- Humanity's Last Exam6.2%
An evaluation missing from this list was not run for this model — it is not a zero.
View as table
| Evaluation | Score |
|---|---|
| MMLU-Pro | 86.0% |
| GPQA Diamond | 70.1% |
| LiveCodeBench | 54.2% |
| IFBench | 43.3% |
| SciCode | 40.9% |
| AA-LCR (long context) | 40.0% |
| AIME 2025 | 36.3% |
| Humanity's Last Exam | 6.2% |
Against its peers
Intelligence Index · this model highlighted, nearest peers in grey
- o4 Mini HighOpenAI26.1
- o4 MiniOpenAI26.1
- Claude Opus 4Anthropic26.0
- Claude Sonnet 4Anthropic26.0
- Gemini 2.5 ProGoogle25.9
- GPT-5 MiniOpenAI25.8
- Gemini 3.1 Flash LiteGoogle25.6
- 25.6
Peers are the models sitting closest on the Intelligence Index — the set you would realistically choose between.
View as table
| Model | Intelligence |
|---|---|
| o4 Mini High | 26.1 |
| o4 Mini | 26.1 |
| Claude Opus 4 | 26.0 |
| Claude Sonnet 4 | 26.0 |
| Gemini 2.5 Pro | 25.9 |
| GPT-5 Mini | 25.8 |
| Gemini 3.1 Flash Lite | 25.6 |
| Gemini 3.1 Flash Lite Preview | 25.6 |
Percentile among all indexed models
Pricing
What it costs to run
List prices per million tokens, plus what one representative task works out to.
List price
- Input / 1M tokens
- $15
- Output / 1M tokens
- $75
- Cached input / 1M
- $1.5
- Blended 3:1
- $30
One task, estimated
$6.75
- Input tokens
- 50,000
- Output tokens
- 80,000
- Profile
- Reasoning
Estimated from list pricing: 50K input tokens plus 80K output tokens for reasoning models (25K for non-reasoning).
Serving providers
Speed and latency figures are medians across these providers, so a widely-served model reports a blend rather than any single endpoint.
Arena
Head-to-head generation quality
Elo from pairwise judgements, broken out by the kind of thing the model was asked to build.
- Overall Elo
- 1,212
- Win rate
- 59.9%
- Strongest at
- Game development
- Tournaments
- 1,652
Elo by category
Dot position on a 1,150–1,225 scale · Elo has no meaningful zero
- 1,212
- 1,194
- 1,188
- 1,187
- 1,174
- 1,173
Agent categories (full-stack apps, mobile apps) are only scored for models tested in agent harnesses.
View as table
| Category | Elo | Win rate |
|---|---|---|
| Game development | 1212 | 59.9% |
| 3D scenes | 1194 | 57.7% |
| UI components | 1188 | 59.2% |
| Websites | 1187 | 54.6% |
| SVG | 1174 | 57.7% |
| Data visualisation | 1173 | 57.9% |