# Claude Haiku 4.5 Thinking benchmark scores

Anthropic · model · https://benchgap.net/model/claude-haiku-4-5-thinking · data retrieved 2026-11-22

As of 2026-11-22, Claude Haiku 4.5 Thinking (Anthropic) has measured scores on 3 benchmarks and estimated scores on 21 more.

| Benchmark | Score | Source |
|---|---|---|
| [AA-SciCode](https://benchgap.net/b/aa-sci-code/current.md) | 47.1% | estimated ± 3.8 pp, high confidence |
| [AA Coding Index](https://benchgap.net/b/aa-coding-index/current.md) | 45.4% | estimated ± 6.9 pp, medium confidence |
| [AA-MMMU-Pro](https://benchgap.net/b/aa-mmmu-pro/current.md) | 65.0% | estimated ± 6.1 pp, medium confidence |
| [SWE-bench Verified](https://benchgap.net/b/swe-bench-verified/current.md) | 73.3% | estimated ± 4.5 pp, high confidence |
| [SWE-bench Pro](https://benchgap.net/b/swe-bench-pro/current.md) | 53.9% | estimated ± 2.9 pp, medium confidence |
| [Mystery Game Puzzles](https://benchgap.net/b/mysterygamepuzzles/current.md) | 14.5% | estimated ± 4.7 pp, medium confidence |
| [OfficeQA Pro](https://benchgap.net/b/officeqapro/current.md) | 48.5% | estimated ± 7.0 pp, low confidence |
| [ParseBench-1](https://benchgap.net/b/parse-bench-llama-index/current.md) | 51.2% | measured |
| [Vals SWE-bench](https://benchgap.net/b/valsswebench/current.md) | 69.1% | estimated ± 7.1 pp, medium confidence |
| [PostTrainBench v1.1](https://benchgap.net/b/posttrainbench%23posttrainbench-v1-1/current.md) | 19.9% | estimated ± 10.5 pp, low confidence |
| [SWE-bench Multimodal](https://benchgap.net/b/swe-bench-multimodal/current.md) | 0.1% | estimated ± 5.6 pp, low confidence |
| [CompileBench](https://benchgap.net/b/compilebench/current.md) | 73.3% | measured |
| [Vibe Code Bench](https://benchgap.net/b/vibecodebench/current.md) | 11.4% | measured |
| [ParseBench-2](https://benchgap.net/b/parsebench/current.md) | 61.3% | estimated ± 4.3 pp, high confidence |
| [CharXiv](https://benchgap.net/b/charxiv/current.md) | 80.5% | estimated ± 5.6 pp, low confidence |
| [CharXiv w/o tools](https://benchgap.net/b/charxivnotools/current.md) | 71.1% | estimated ± 1.8 pp, low confidence |
| [MMMU-Pro](https://benchgap.net/b/mmmu-pro/current.md) | 71.9% | estimated ± 6.9 pp, medium confidence |
| [PostTrainBench v1](https://benchgap.net/b/posttrainbench%23posttrainbench-v1/current.md) | 19.0% | estimated ± 3.1 pp, low confidence |
| [ScreenSpot Pro](https://benchgap.net/b/screenspot-pro/current.md) | 74.3% | estimated ± 10.1 pp, low confidence |
| [AA LiveCodeBench](https://benchgap.net/b/aa-live-code-bench/current.md) | 85.1% | estimated ± 3.4 pp, low confidence |
| [React Native Evals](https://benchgap.net/b/reactnativeevals/current.md) | 57.0% | estimated ± 3.0 pp, low confidence |
| [SWE-Rebench](https://benchgap.net/b/swe-rebench/current.md) | 2.2% | estimated ± 6.5 pp, low confidence |
| [PostTrainBench](https://benchgap.net/b/post-train-bench12/current.md) | 17.0% | estimated ± 8.3 pp, low confidence |
| [MMMU-Pro w/ Python](https://benchgap.net/b/mmmupropython/current.md) | 75.2% | estimated ± 2.7 pp, medium confidence |
