# Claude Haiku 4.5 vs Claude Sonnet 4.6

> Side-by-side benchmark comparison for Claude Haiku 4.5 and Claude Sonnet 4.6 across agentic, coding, multimodal, knowledge, reasoning, multilingual, and math tasks.

- Canonical page: https://benchlm.ai/compare/claude-haiku-4-5-vs-claude-sonnet-4-6
- Last updated: October 17, 2026

- Shared sourced benchmarks: 8
- HTML indexing: indexable
- Ranking lane: BenchAlign v5.8

## Quick Verdict

Claude Sonnet 4.6 leads by point estimate. Conditional score ranges do not establish rank confidence. Choose using the category evidence and your own workload tests.

## Summary

- Claude Sonnet 4.6 has the higher overall point estimate 55.89 to 41.16. Conditional score ranges do not establish rank confidence.
- The clearest category separation is in coding, where the averages are 16.8 for Claude Haiku 4.5 and 44.7 for Claude Sonnet 4.6.
- The biggest single benchmark swing is LiveCodeBench (Vals) in Coding, with scores of 41.2% and 82.1%.
- Claude Haiku 4.5 is the cheaper option on output tokens, which matters if you expect large responses or heavy interactive use.

## Model Snapshot

| Property | Claude Haiku 4.5 | Claude Sonnet 4.6 |
|----------|----------|----------|
| Creator | Anthropic | Anthropic |
| Type | Proprietary | Proprietary |
| Reasoning | Non-Reasoning | Non-Reasoning |
| Context | 200K | 200K |
| Overall Score | 41.16 | 55.89 |
| Benchmarks Covered | 9 | 28 |
| Pricing (input/output) | $1.00 / $5.00 | $3.00 / $15.00 |

## Category Breakdown

### Agentic

- Winner: Claude Sonnet 4.6
- Claude Haiku 4.5 public-lane score: 21.4 (Supported · #107/128)
- Claude Sonnet 4.6 public-lane score: 46.3 (Supported · #55/128)

| Benchmark | Claude Haiku 4.5 | Claude Sonnet 4.6 | Winner |
|-----------|-----------|-----------|--------|
| JobBench | 16.0% | 36.9% | Claude Sonnet 4.6 |
| Terminal-Bench 2.1 (Vals) | 43.8% | 57.3% | Claude Sonnet 4.6 |
| MazeBench | 0% | 0% | Tie |
| Terminal-Bench 2.0 | Coming soon | 59.1% | Coming soon |
| OSWorld-Verified | Coming soon | 72.1% | Coming soon |
| Claw-Eval | Coming soon | 67.8% | Coming soon |
| CyberGym | Coming soon | 65.2% | Coming soon |
| Gert Labs | Coming soon | 62.92% | Coming soon |
| OSWorld 2.0 | Coming soon | 8.3% | Coming soon |
| ApprenticeBench | Coming soon | 2% | Coming soon |

### Coding

- Winner: Claude Sonnet 4.6
- Claude Haiku 4.5 public-lane score: 16.8 (Supported · #126/150)
- Claude Sonnet 4.6 public-lane score: 44.7 (Supported · #54/150)

| Benchmark | Claude Haiku 4.5 | Claude Sonnet 4.6 | Winner |
|-----------|-----------|-----------|--------|
| SWE-bench Verified | 73.3% | 79.6% | Claude Sonnet 4.6 |
| VulcanBench v3 | 76.2% | Coming soon | Coming soon |
| LiveCodeBench (Vals) | 41.2% | 82.1% | Claude Sonnet 4.6 |
| SWE-bench (Vals) | 66.6% | 77.4% | Claude Sonnet 4.6 |
| SWE-Rebench | Coming soon | 60.7% | Coming soon |
| React Native Evals | Coming soon | 80.6% | Coming soon |
| Terminal-Bench 2.0 | Coming soon | 59.1% | Coming soon |
| Vibe Code Bench | Coming soon | 51.48% | Coming soon |
| cursorBench31 | Coming soon | 48.8% | Coming soon |
| FrontierCode 1.1 Main | Coming soon | 24.3% | Coming soon |

### Reasoning

- Winner: Not comparable
- Claude Haiku 4.5 public-lane score: Coming soon
- Claude Sonnet 4.6 public-lane score: 62.8 (Unranked · 4 rankable rows)

| Benchmark | Claude Haiku 4.5 | Claude Sonnet 4.6 | Winner |
|-----------|-----------|-----------|--------|
| ARC-AGI-2 | Coming soon | 58.3% | Coming soon |
| ARC-AGI-1 | Coming soon | 86.00% | Coming soon |

### Multimodal & Grounded

- Winner: Not comparable
- Claude Haiku 4.5 public-lane score: Coming soon
- Claude Sonnet 4.6 public-lane score: 58.7 (#45/54)

| Benchmark | Claude Haiku 4.5 | Claude Sonnet 4.6 | Winner |
|-----------|-----------|-----------|--------|
| CharXiv | Coming soon | 77.4% | Coming soon |

### Knowledge

- Winner: Directional only
- Claude Haiku 4.5 public-lane score: 33.3 (Estimated · #141/184)
- Claude Sonnet 4.6 public-lane score: 55.4 (Supported · #58/184)

| Benchmark | Claude Haiku 4.5 | Claude Sonnet 4.6 | Winner |
|-----------|-----------|-----------|--------|
| GPQA Diamond (Vals) | 72.2% | 85.6% | Claude Sonnet 4.6 |
| MMLU-Pro (Vals) | 78.7% | 87.3% | Claude Sonnet 4.6 |
| GPQA | Coming soon | 89.9% | Coming soon |
| SuperGPQA | Coming soon | 95% | Coming soon |
| MMLU-Pro | Coming soon | 79.2% | Coming soon |
| HLE | Coming soon | 49% | Coming soon |

### Instruction Following

- Winner: Not comparable
- Claude Haiku 4.5 public-lane score: Coming soon
- Claude Sonnet 4.6 public-lane score: 48 (#99/148)

## FAQ

### Which is better overall, Claude Haiku 4.5 or Claude Sonnet 4.6?

Claude Sonnet 4.6 has the higher overall point estimate. Conditional score ranges do not establish rank confidence.

### Where is the biggest gap between Claude Haiku 4.5 and Claude Sonnet 4.6?

The widest category gap is in coding, where the averages are 16.8 for Claude Haiku 4.5 and 44.7 for Claude Sonnet 4.6.

### How many benchmarks does Claude Haiku 4.5 cover on BenchLM?

Claude Haiku 4.5 currently has 9 sourced benchmark scores on BenchLM.

### How many benchmarks does Claude Sonnet 4.6 cover on BenchLM?

Claude Sonnet 4.6 currently has 28 sourced benchmark scores on BenchLM.

## Related Comparisons

- [Claude Haiku 4.5 vs Claude Haiku 4.5 Thinking](/compare/claude-haiku-4-5-vs-claude-haiku-4-5-thinking)
- [Claude Sonnet 4.6 vs Claude Haiku 4.5 Thinking](/compare/claude-haiku-4-5-thinking-vs-claude-sonnet-4-6)
- [Claude Haiku 4.5 vs Claude Opus 5.5](/compare/claude-haiku-4-5-vs-claude-opus-5-5)
- [Claude Sonnet 4.6 vs Claude Opus 5.5](/compare/claude-opus-5-5-vs-claude-sonnet-4-6)

## Explore More

- [Claude Haiku 4.5 profile](/models/claude-haiku-4-5)
- [Claude Sonnet 4.6 profile](/models/claude-sonnet-4-6)
- [Compare Pricing](/llm-pricing)
- [Alternative Finder](/tools/alternative-finder)
- [LLM Selector](/tools/llm-selector)
- [Overall Rankings](/best/overall)