Claude Sonnet 4.5 (batch)
Anthropic · released Sep 29, 2025
Claude Sonnet 4.5 is Anthropic’s most advanced Sonnet model to date, optimized for real-world agents and coding workflows. It delivers state-of-the-art performance on coding benchmarks such as SWE-bench Verified, with...
Specification
- Context window
- 1M
- Max output
- 64K
- Knowledge cutoff
- Jan 31, 2025
- Parameters
- Undisclosed
- Licence
- Proprietary
- Serving providers
- 1
- Moderated
- Yes
Intelligence
29.9
73th percentile
Coding
—
Coding Index
Agentic
26.4
Agentic Index
Output speed
—
Median across providers
Latency
—
Time to first token
Cost per task
$0.68
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 Diamond72.7%
- τ²-bench (Telecom)70.5%
- LiveCodeBench59.0%
- AA-LCR (long context)52.3%
- SciCode42.8%
- IFBench42.7%
- AIME 202537.0%
- Terminal-Bench Hard28.8%
- Humanity's Last Exam7.2%
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 | 72.7% |
| τ²-bench (Telecom) | 70.5% |
| LiveCodeBench | 59.0% |
| AA-LCR (long context) | 52.3% |
| SciCode | 42.8% |
| IFBench | 42.7% |
| AIME 2025 | 37.0% |
| Terminal-Bench Hard | 28.8% |
| Humanity's Last Exam | 7.2% |
Against its peers
Intelligence Index · this model highlighted, nearest peers in grey
- Ring-2.6-1TInclusionAI31.1
- Step 3.7 FlashStepFun30.9
- Mistral Medium 3.5Mistral AI30.4
- Claude Sonnet 4.5 (batch)Anthropic29.9
- Qwen3.5-35B-A3BQwen29.9
- Claude Sonnet 4.5Anthropic29.9
- MiniMax M2MiniMax28.9
- Claude Opus 4.1Anthropic28.8
Peers are the models sitting closest on the Intelligence Index — the set you would realistically choose between.
View as table
| Model | Intelligence |
|---|---|
| Ring-2.6-1T | 31.1 |
| Step 3.7 Flash | 30.9 |
| Mistral Medium 3.5 | 30.4 |
| Claude Sonnet 4.5 (batch) | 29.9 |
| Qwen3.5-35B-A3B | 29.9 |
| Claude Sonnet 4.5 | 29.9 |
| MiniMax M2 | 28.9 |
| Claude Opus 4.1 | 28.8 |
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
- $1.5
- Output / 1M tokens
- $7.5
- Cached input / 1M
- $0.15
- Blended 3:1
- $3
One task, estimated
$0.68
- 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
- 1,213
- Win rate
- 51.1%
- Strongest at
- 3D scenes
- Tournaments
- 7,252
Elo by category
Dot position on a 1,150–1,225 scale · Elo has no meaningful zero
- 1,213
- 1,212
- 1,208
- 1,207
- 1,194
- 1,162
Agent categories (full-stack apps, mobile apps) are only scored for models tested in agent harnesses.
View as table
| Category | Elo | Win rate |
|---|---|---|
| 3D scenes | 1213 | 51.1% |
| Websites | 1212 | 52.0% |
| UI components | 1208 | 49.5% |
| Game development | 1207 | 51.2% |
| Data visualisation | 1194 | 47.3% |
| SVG | 1162 | 52.2% |