Claude Haiku 4.5
| Released | 2025-10-15 |
|---|---|
| Price per 1M | $1.00 in · $5.00 out |
| Overall score | 46.7 · rank 142 |
| Context window | 200,000 tokens |
| Provider | Anthropic |
| Weights | Proprietary |
Scores
| Dimension | Score | Rank | Benchmarks |
|---|---|---|---|
| Overall: Balanced | 46.7 | 142 | 27 |
| Overall: Coding | 47.4 | 129 | 17 |
| Overall: Reasoning | 46.7 | 142 | 12 |
| Overall: Agentic work | 49.0 | 119 | 15 |
| Overall: Frontend | 42.6 | 160 | 7 |
| Coding | 47.9 | 134 | 4 |
| Agents | 49.9 | 107 | 6 |
| Reasoning | 47.7 | 130 | 5 |
| Math | 48.3 | 128 | 4 |
| Knowledge | 41.9 | 136 | 3 |
| Frontend | 42.2 | 106 | 2 |
| General | 50.1 | 109 | 2 |
| Multimodal | 29.8 | 118 | 1 |
| Value vs price | −3.1 | 149 |
Configurations
| Configuration | Balanced | Benchmarks |
|---|---|---|
| Unspecified setting | 46.2 ±1.5 | 16 |
| Low | 50.0 ±7.9 | 1 |
| Thinking | 46.2 ±3.1 | 19 |
Evidence
Coding 47.9 ±6.8
| Benchmark | Result | Source | Date |
|---|---|---|---|
| SWE-bench Verified (Vals run) | 66.6% | Vals AI | 2026-09-01 |
| SciCode | 42.3% | Epoch AI | 2026-11-11 |
| LiveCodeBench (Vals run) | 41.2% | Vals AI | 2026-09-01 |
| ProgramBench | 0.0% | Vals AI | 2026-11-09 |
Agents 49.9 ±1.8
| Benchmark | Result | Source | Date |
|---|---|---|---|
| BFCL v4 | 68.7% | BFCL | 2026-04-12 |
| Terminal-Bench 2.0 | 35.5% | Epoch AI | 2025-10-15 |
| Tax Agent Bench | 5.9% | Vals AI | 2026-11-09 |
| Terminal-Bench 2.1 (Vals run) | 43.8% | Vals AI | 2026-10-01 |
| DeepResearch Bench | 45.5% | Epoch AI | 2026-11-11 |
| BALROG | 31.2% | Epoch AI | 2026-11-11 |
| Vending-Bench 2 | $459 | Epoch AI | 2026-11-11 |
Reasoning 47.7 ±3.1
| Benchmark | Result | Source | Date |
|---|---|---|---|
| ARC-AGI-2 | 4.0% | Epoch AI | 2026-11-11 |
| ARC-AGI-1 | 47.7% | Epoch AI | 2026-11-11 |
| CritPt | 0.0% | Epoch AI | 2026-11-11 |
| GPQA Diamond | 71.2% | Epoch AI | 2025-10-22 |
| GPQA Diamond (Vals run) | 72.2% | Vals AI | 2026-09-01 |
| Chess Puzzles | 8.0% | Epoch AI | 2026-10-13 |
Math 48.3 ±1.6
| Benchmark | Result | Source | Date |
|---|---|---|---|
| MATH Level 5 | 96.4% | Epoch AI | 2025-10-22 |
| FrontierMath Tiers 1-3 (2025 set) | 5.9% | Epoch AI | 2025-10-22 |
| FrontierMath Tier 4 (2025 set) | 2.1% | Epoch AI | 2025-10-22 |
| OTIS Mock AIME 2024-2025 | 66.7% | Epoch AI | 2025-10-22 |
Knowledge 41.9 ±3.6
| Benchmark | Result | Source | Date |
|---|---|---|---|
| SimpleQA Verified | 13.2% | Epoch AI | 2026-08-10 |
| MMLU-Pro (Vals run) | 78.7% | Vals AI | 2026-09-01 |
| Legal Research Bench | 10.6% | Vals AI | 2026-11-09 |
Frontend 42.2 ±0.7
| Benchmark | Result | Source | Date |
|---|---|---|---|
| Vibe Code Bench v1.1 | 11.4% | Vals AI | 2026-11-09 |
| LMArena WebDev | 1,330 | LMArena | 2026-11-10 |
General 50.1 ±2.6
| Benchmark | Result | Source | Date |
|---|---|---|---|
| EQ-Bench 4 | 1,053 | EQ-Bench | 2026-10-24 |
| LMArena Text | 1,414 | LMArena | 2026-11-09 |
Multimodal 29.8 ±2.7
| Benchmark | Result | Source | Date |
|---|---|---|---|
| MMMU-Pro (Vals run) | 46.1% | Vals AI | 2026-09-01 |
Results not used in the score
| Benchmark | Result | Source | Why not used |
|---|---|---|---|
| ARC-AGI-1 | 14.3% | Epoch AI | A stronger configuration result was used |
| ARC-AGI-2 | 1.3% | Epoch AI | A stronger configuration result was used |
| BALROG | 31.2% | Epoch AI | A stronger configuration result was used |
| BFCL v4 | 25.3% | BFCL | A stronger configuration result was used |
| FrontierMath Tiers 1-3 (2025 set) | 4.1% | Epoch AI | A stronger configuration result was used |
| GPQA Diamond | 60.5% | Epoch AI | A stronger configuration result was used |
| MATH Level 5 | 86.9% | Epoch AI | A stronger configuration result was used |
| OTIS Mock AIME 2024-2025 | 35.8% | Epoch AI | A stronger configuration result was used |
| SimpleQA Verified | 12.6% | Epoch AI | A stronger configuration result was used |
| Terminal-Bench 2.0 | 29.8% | Epoch AI | A stronger configuration result was used |
| Terminal-Bench 2.0 | 28.3% | Epoch AI | A stronger configuration result was used |
| Terminal-Bench 2.0 | 27.5% | Epoch AI | A stronger configuration result was used |
| Terminal-Bench 2.0 | 13.9% | Epoch AI | A stronger configuration result was used |
Names matched to this model
claude-haiku-4-5-20251001