Model comparison
Claude 3 Haiku vs Claude Haiku 4.5
Claude Haiku 4.5 is the stronger model overall, scoring 39.5 to 25.9 on the Noometry Index.
Last verified . 25 shared benchmarks.
Summary
- They share 25 benchmarks with published results for both. Claude 3 Haiku scores higher in 1 category and Claude Haiku 4.5 in 8 categories; 9 gaps are clear of the uncertainty.
- The widest gap is in math, where Claude Haiku 4.5 leads 44.9 to 9.8.
- The biggest single-benchmark swing is MATH Level 5: 14.9% for Claude 3 Haiku and 96.4% for Claude Haiku 4.5.
Side by side
| Claude 3 Haiku | Claude Haiku 4.5 | |
|---|---|---|
| Provider | Anthropic | Anthropic |
| Noometry Index | 25.9 | 39.5 |
| Released | 2024-03-07 | 2025-10-15 |
| Weights | Proprietary | Proprietary |
| Context window | — | 200K |
| Max output | — | 64K |
| Input $ / M tokens | — | $1 |
| Output $ / M tokens | — | $5 |
| Results tracked | 37 | 53 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Claude Haiku 4.5 leads
Claude 3 Haiku: 26.4 (#325), Claude Haiku 4.5: 44.0 (#78)
| Benchmark | Claude 3 Haiku | Claude Haiku 4.5 |
|---|---|---|
| WeirdML | 9.8% | 45.4% |
| LMArena Coding | 1199 | 1453 |
| SWE-bench Verified (bash only) | — | 66.6% |
| LMArena WebDev | — | 1330 |
| SWE-bench Multilingual | — | 64.7% |
| SciCode | — | 43.3% |
| BigCodeBench Instruct | 39.4% | — |
| BigCodeBench Complete | 50.1% | — |
| CadEval | 12% | — |
| ALE-Bench | — | 653.48 |
| HumanEval+ | 68.9% | — |
| MBPP+ | 68.8% | — |
Agentic & Tool Use Not comparable
Claude 3 Haiku: —, Claude Haiku 4.5: 33.6 (#52)
| Benchmark | Claude 3 Haiku | Claude Haiku 4.5 |
|---|---|---|
| Terminal-Bench | — | 35.5% |
| Berkeley Function Calling Leaderboard | — | 68.7% |
| DeepResearch Bench | — | 45.5% |
| BALROG | — | 31.2% |
| ExploitBench | — | 13.7% |
| Vending-Bench 2 | — | 458.89 |
Reasoning Claude 3 Haiku leads
Claude 3 Haiku: 16.3 (#307), Claude Haiku 4.5: 15.1 (#320)
| Benchmark | Claude 3 Haiku | Claude Haiku 4.5 |
|---|---|---|
| LMArena Hard Prompts | 1174 | 1420 |
| DTBench | 50.1% | 73.6% |
| LMCA | 8.8% | 30.9% |
| Epoch Capabilities Index | 118.35 | 142.41 |
| ForecastBench | 53.2 | 61.4 |
| ARC-AGI-2 | — | 4% |
| Kagi LLM Benchmark | 34.2% | — |
| NYT Connections (extended) | — | 14.3% |
| ARC-AGI-1 | — | 47.7% |
| CritPt | — | 0% |
| Chess Puzzles | — | 8% |
| WinoGrande | 74.2% | — |
Math Claude Haiku 4.5 leads
Claude 3 Haiku: 9.8 (#319), Claude Haiku 4.5: 44.9 (#78)
| Benchmark | Claude 3 Haiku | Claude Haiku 4.5 |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 1.8% | 66.7% |
| LMArena Math | 1188 | 1396 |
| MATH Level 5 | 14.9% | 96.4% |
| Omni-MATH | — | 56.1% |
| FrontierMath (Feb 2025 set) | — | 5.9% |
| FrontierMath Tier 4 (v1) | — | 2.1% |
Knowledge Claude Haiku 4.5 leads
Claude 3 Haiku: 17.3 (#285), Claude Haiku 4.5: 37.7 (#153)
| Benchmark | Claude 3 Haiku | Claude Haiku 4.5 |
|---|---|---|
| GPQA Diamond | 36.3% | 71.2% |
| LMArena Expert | 1148 | 1442 |
| SimpleQA Verified | — | 13.2% |
| MMLU-Pro | — | 77.7% |
| Confabulations | 34.2% | — |
| Vectara Hallucination Rate | — | 9.8% |
| GPQA (HELM) | — | 60.5% |
| MMLU | 73.8% | — |
Multimodal Claude Haiku 4.5 leads
Claude 3 Haiku: 23.6 (#128), Claude Haiku 4.5: 26.8 (#118)
| Benchmark | Claude 3 Haiku | Claude Haiku 4.5 |
|---|---|---|
| LMArena Vision | 950 | — |
| Blueprint-Bench 2 | — | 0% |
| LMArena Document | — | 1420 |
| ScienceQA | 72% | — |
Multilingual Claude Haiku 4.5 leads
Claude 3 Haiku: 36.0 (#243), Claude Haiku 4.5: 49.9 (#129)
| Benchmark | Claude 3 Haiku | Claude Haiku 4.5 |
|---|---|---|
| LMArena Non-English | 1178 | 1377 |
| LMArena Chinese | 1155 | 1417 |
| LMArena French | 1195 | 1408 |
| LMArena German | 1174 | 1375 |
| LMArena Japanese | 1102 | 1339 |
| LMArena Korean | 1109 | 1347 |
| LMArena Russian | 1204 | 1381 |
| LMArena Spanish | 1166 | 1420 |
Instruction Following Claude Haiku 4.5 leads
Claude 3 Haiku: 61.3 (#247), Claude Haiku 4.5: 71.4 (#149)
| Benchmark | Claude 3 Haiku | Claude Haiku 4.5 |
|---|---|---|
| LMArena Instruction Following | 1173 | 1414 |
| IFEval | — | 80.1% |
Long Context Claude Haiku 4.5 leads
Claude 3 Haiku: 36.1 (#237), Claude Haiku 4.5: 43.6 (#92)
| Benchmark | Claude 3 Haiku | Claude Haiku 4.5 |
|---|---|---|
| LMArena Longer Query | 1190 | 1427 |
Writing & Preference Claude Haiku 4.5 leads
Claude 3 Haiku: 29.7 (#291), Claude Haiku 4.5: 57.9 (#123)
| Benchmark | Claude 3 Haiku | Claude Haiku 4.5 |
|---|---|---|
| LMArena Text | 1195 | 1396 |
| LMArena Creative Writing | 1157 | 1372 |
| LMArena Multi-Turn | 1190 | 1409 |
| EQ-Bench Creative Writing | 717 | — |
| WildBench | — | 83.9% |
| EQ-Bench 4 | — | 1064 |
Frequently asked questions
Is Claude 3 Haiku better than Claude Haiku 4.5?
Claude Haiku 4.5 is the stronger model overall, scoring 39.5 to 25.9 on the Noometry Index.
Is Claude 3 Haiku or Claude Haiku 4.5 better for coding?
Claude Haiku 4.5 scores higher on coding benchmarks: 44.0 versus 26.4 in the Noometry coding category.
How many benchmarks do Claude 3 Haiku and Claude Haiku 4.5 share?
25 benchmarks have published results for both models. Claude 3 Haiku has 37 scored results on Noometry and Claude Haiku 4.5 has 53.