# Claude Sonnet 4.6 Benchmark Scores & Performance

> Claude Sonnet 4.6 by Anthropic scores 55.89/100 overall, ranking #64 out of 954 AI models.

Canonical page: https://benchlm.ai/models/claude-sonnet-4-6

Last updated: October 17, 2026

## Model Details

| Property | Value |
|----------|-------|
| Creator | Anthropic |
| Source Type | Proprietary |
| Reasoning Type | Non-Reasoning |
| Context Window | 200K |
| Overall Score | 55.89/100 |
| Overall Rank | #64 of 954 |

## Family & Coverage

- Family: Claude Sonnet 4.6
- Variant: base
- Benchmarks covered: 40 of 639
- Coverage note: BenchLM currently has partial benchmark coverage for this model, so the overall score is conservative.

## Agentic Benchmarks

| Benchmark | Score | Source |
|-----------|-------|--------|
| [Terminal-Bench 2.0](/benchmarks/terminal-bench-2) | 59.1% | [Anthropic: Claude Sonnet 4.6 system card](https://www-cdn.anthropic.com/78073f739564e986ff3e28522761a7a0b4484f84.pdf) |
| [OSWorld-Verified](/benchmarks/osworld-verified) | 72.1% | [OSWorld-Verified leaderboard](https://os-world.github.io/) |
| [Claw-Eval](/benchmarks/claw-eval) | 67.8% | [Claw-Eval leaderboard](https://claw-eval.github.io/) |
| [CyberGym](/benchmarks/cybergym) | 65.2% | [CyberGym leaderboard](https://cybergym.io/) |
| [τ²-bench results](/benchmarks/tau2-bench) | 79.5% | [Artificial Analysis model benchmarks](https://artificialanalysis.ai/models/claude-sonnet-4-6) |
| [Gert Labs](/benchmarks/gertlabs) | 62.92% | [Gert Labs rankings](https://gertlabs.com/rankings) |
| [OSWorld 2.0](/benchmarks/osworld2) | 8.3% | [OSWorld 2.0 paper](https://arxiv.org/abs/2606.29537) |
| [JobBench](/benchmarks/jobbench) | 36.9% | [JobBench paper](https://arxiv.org/abs/2605.26329) |
| [Terminal-Bench 2.1 (Vals)](/benchmarks/terminalbench21) | 57.3% | [Vals AI: Terminal-Bench 2.1 leaderboard](https://www.vals.ai/models/anthropic_claude-sonnet-4-6) |
| [ApprenticeBench](/benchmarks/apprenticebench) | 2% | [NeoCognition: ApprenticeBench GUI results](https://neocognition.io/blog/apprentice-bench/#are-frontier-agents-ready-for-the-job) |
| [MazeBench](/benchmarks/mazebench) | 0% | [MazeBench AI leaderboard (ASCII mode)](https://mazebench.com/leaderboard/ai?mode=ascii&tools=0&metric=gems&run=2026-07-21T01-21-49-520-821873) |

## Coding Benchmarks

| Benchmark | Score | Source |
|-----------|-------|--------|
| [SWE-bench Verified](/benchmarks/swe-bench-verified) | 79.6% | [Anthropic: Claude Sonnet 4.6 system card](https://www-cdn.anthropic.com/78073f739564e986ff3e28522761a7a0b4484f84.pdf) |
| [SWE-Rebench](/benchmarks/swe-rebench) | 60.7% | [SWE-Rebench leaderboard](https://swe-rebench.com/?insight=sep_2025) |
| [React Native Evals](/benchmarks/reactnativeevals) | 80.6% | [React Native Evals leaderboard](https://rn-evals.vercel.app/) |
| [Terminal-Bench 2.0](/benchmarks/terminal-bench-2) | 59.1% | [Anthropic: Claude Sonnet 4.6 system card](https://www-cdn.anthropic.com/78073f739564e986ff3e28522761a7a0b4484f84.pdf) |
| [Vibe Code Bench](/benchmarks/vibecodebench) | 51.48% | [Vals AI: Vibe Code Bench v1.1](https://www.vals.ai/benchmarks/vibe-code) |
| [cursorBench31](/benchmarks/cursorbench31) | 48.8% | [Cursor evals: CursorBench 3.1](https://cursor.com/evals) |
| [FrontierCode 1.1 Main](/benchmarks/frontiercode) | 24.3% | [Cognition: FrontierCode 1.1](https://cognition.com/blog/frontier-code-1.1) |
| [LiveCodeBench (Vals)](/benchmarks/valslivecodebench) | 82.1% | [Vals AI: LiveCodeBench leaderboard](https://www.vals.ai/models/anthropic_claude-sonnet-4-6) |
| [SWE-bench (Vals)](/benchmarks/valsswebench) | 77.4% | [Vals AI: SWE-bench leaderboard](https://www.vals.ai/models/anthropic_claude-sonnet-4-6) |

## Multimodal & Grounded Benchmarks

| Benchmark | Score | Source |
|-----------|-------|--------|
| [CharXiv](/benchmarks/charxiv) | 77.4% | [Anthropic: Claude Sonnet 4.6 system card](https://www-cdn.anthropic.com/bbd8ef16d70b7a1665f14f306ee88b53f686aa75.pdf) |
| [AA-MMMU-Pro](/benchmarks/mmmu-pro) | 70.6% | [Artificial Analysis model benchmarks](https://artificialanalysis.ai/models/claude-sonnet-4-6) |
| [Design Arena Website](/benchmarks/designarenawebsite) | 1286 | [OpenRouter model benchmarks](https://openrouter.ai/anthropic/claude-sonnet-4.6/benchmarks) |

## Reasoning Benchmarks

| Benchmark | Score | Source |
|-----------|-------|--------|
| [ARC-AGI-2](/benchmarks/arc-agi-2) | 58.3% | [ARC Prize: Claude Sonnet 4.6 (Max) verified results](https://arcprize.org/leaderboard) |
| [AA-LCR](/benchmarks/lcr) | 68.3% | [Artificial Analysis model benchmarks](https://artificialanalysis.ai/models/claude-sonnet-4-6) |
| [CritPt](/benchmarks/critpt) | 0.9% | [Artificial Analysis model benchmarks](https://artificialanalysis.ai/models/claude-sonnet-4-6) |
| [ARC-AGI-1](/benchmarks/arcagi1) | 86.00% | [ARC Prize: Claude Sonnet 4.6 (Max) verified results](https://arcprize.org/leaderboard) |

## Knowledge Benchmarks

| Benchmark | Score | Source |
|-----------|-------|--------|
| [GPQA](/benchmarks/gpqa) | 89.9% | [Anthropic: Claude Sonnet 4.6 system card](https://www-cdn.anthropic.com/78073f739564e986ff3e28522761a7a0b4484f84.pdf) |
| [SuperGPQA](/benchmarks/supergpqa) | 95% | [Qwen3.6-Plus comparison table](https://qwen.ai/blog?id=qwen3.6) |
| [MMLU-Pro](/benchmarks/mmlu-pro) | 79.2% | [Qwen3.6-Plus comparison table](https://qwen.ai/blog?id=qwen3.6) |
| [HLE](/benchmarks/hle) | 49% | [Anthropic: Claude Sonnet 4.6 system card](https://www-cdn.anthropic.com/78073f739564e986ff3e28522761a7a0b4484f84.pdf) |
| [Artificial Analysis Intelligence Index](/benchmarks/artificialanalysis) | 24.7% | [Artificial Analysis model benchmarks](https://artificialanalysis.ai/models/claude-sonnet-4-6) |
| [AA-GPQA Diamond](/benchmarks/gpqa-diamond) | 79.9% | [Artificial Analysis model benchmarks](https://artificialanalysis.ai/models/claude-sonnet-4-6) |
| [AA-HLE](/benchmarks/aahle) | 13.3% | [Artificial Analysis model benchmarks](https://artificialanalysis.ai/models/claude-sonnet-4-6) |
| [AA-Omniscience Index](/benchmarks/aaomniscienceindex) | -3.5% | [Artificial Analysis model benchmarks](https://artificialanalysis.ai/models/claude-sonnet-4-6) |
| [AA-Omniscience Accuracy](/benchmarks/omniscienceaccuracy) | 38.6% | [Artificial Analysis model benchmarks](https://artificialanalysis.ai/models/claude-sonnet-4-6) |
| [AA-Omniscience Hallucination Rate](/benchmarks/omnisciencehallucinationrate) | 68.5% | [Artificial Analysis model benchmarks](https://artificialanalysis.ai/models/claude-sonnet-4-6) |
| [GPQA Diamond (Vals)](/benchmarks/gpqa-diamond) | 85.6% | [Vals AI: GPQA Diamond leaderboard](https://www.vals.ai/models/anthropic_claude-sonnet-4-6) |
| [MMLU-Pro (Vals)](/benchmarks/mmlu-pro) | 87.3% | [Vals AI: MMLU Pro leaderboard](https://www.vals.ai/models/anthropic_claude-sonnet-4-6) |

## Instruction Following Benchmarks

| Benchmark | Score | Source |
|-----------|-------|--------|
| [AA-IFBench](/benchmarks/ifbench) | 41.2% | [Artificial Analysis model benchmarks](https://artificialanalysis.ai/models/claude-sonnet-4-6) |

## Other Anthropic Models

- [Claude Opus 5.5](/models/claude-opus-5-5) - Score: 86.56
- [Claude Fable 5.1](/models/claude-fable-5-1) - Score: 83.18
- [Claude Sonnet 5.5](/models/claude-sonnet-5-5) - Score: 82.18
- [Claude Fable 5](/models/claude-fable) - Score: 81.27
- [Claude Opus 5](/models/claude-opus-5) - Score: 79.63
- [Claude Haiku 5.5](/models/claude-haiku-5-5) - Score: 70.04
- [Claude Opus 4.8](/models/claude-opus-4-8) - Score: 70.01
- [Claude Opus 4.7 (Adaptive)](/models/claude-opus-4-7-adaptive) - Score: 68.96
- [Claude Opus 4.7](/models/claude-opus-4-7) - Score: 66.74
- [Claude Sonnet 5](/models/claude-sonnet-5) - Score: 64.84
- [Claude Opus 4.6](/models/claude-opus-4-6) - Score: 62.39
- [Claude Opus 4.5 Thinking](/models/claude-opus-4-5-thinking) - Score: 54.1
- [Claude Opus 4.5](/models/claude-opus-4-5) - Score: 54.05
- [Claude Sonnet 4.5](/models/claude-sonnet-4-5) - Score: 47.96
- [Claude Haiku 4.5](/models/claude-haiku-4-5) - Score: 41.16
- [Claude 4.1 Opus](/models/claude-4-1-opus) - Score: 40.05
- [Claude 4.1 Opus Thinking](/models/claude-4-1-opus-thinking) - Score: 39.37
- [Claude 4 Sonnet](/models/claude-4-sonnet) - Score: 38.08
- [Claude 3 Opus](/models/claude-3-opus) - Score: 28.49
- [Claude 3.5 Sonnet](/models/claude-3-5-sonnet) - Score: 27.73
- [Claude 3 Haiku](/models/claude-3-haiku) - Score: 14.19
- [Claude Mythos 5](/models/claude-mythos-5) - Score: not computed
- [Claude 4 Opus Thinking](/models/claude-4-opus-thinking) - Score: not computed
- [Claude Opus 4.6 (Adaptive)](/models/claude-opus-4-6-thinking) - Score: not computed
- [Claude 4 Opus](/models/claude-4-opus) - Score: not computed
- [Claude 3.7 Sonnet Thinking](/models/claude-3-7-sonnet-thinking) - Score: not computed
- [Claude 3.7 Sonnet](/models/claude-3-7-sonnet) - Score: not computed
- [Claude 3.5 Haiku](/models/claude-3-5-haiku) - Score: not computed
- [Claude Sonnet 4.5 Thinking](/models/claude-sonnet-4-5-thinking) - Score: not computed
- [Claude Mythos 5.1](/models/claude-mythos-5-1) - Score: not computed
- [Claude Mythos Preview](/models/claude-mythos-preview) - Score: not computed
- [Claude Haiku 4.5 Thinking](/models/claude-haiku-4-5-thinking) - Score: not computed