Model comparison
Claude 3.5 Haiku vs Qwen2.5 72B Instruct
Qwen2.5 72B Instruct is the stronger model overall, scoring 31.9 to 29.2 on the Noometry Index.
Last verified . 33 shared benchmarks.
Alibaba (Qwen)Summary
- They share 33 benchmarks with published results for both. Claude 3.5 Haiku scores higher in 1 category and Qwen2.5 72B Instruct in 8 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Qwen2.5 72B Instruct leads 27.0 to 18.7.
- The biggest single-benchmark swing is Confabulations: 36.7% for Claude 3.5 Haiku and 19.1% for Qwen2.5 72B Instruct.
- Qwen2.5 72B Instruct has downloadable open weights; the other is API-only.
Side by side
| Claude 3.5 Haiku | Qwen2.5 72B Instruct | |
|---|---|---|
| Provider | Anthropic | Alibaba (Qwen) |
| Noometry Index | 29.2 | 31.9 |
| Released | 2024-10-22 | 2024-09 |
| Weights | Proprietary | Open |
| Context window | — | 131K |
| Max output | — | 8K |
| Input $ / M tokens | — | $1.40 |
| Output $ / M tokens | — | $5.60 |
| Results tracked | 49 | 43 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Too close to call
Claude 3.5 Haiku: 32.9 (#265), Qwen2.5 72B Instruct: 33.2 (#260)
| Benchmark | Claude 3.5 Haiku | Qwen2.5 72B Instruct |
|---|---|---|
| WeirdML | 30.7% | 16% |
| BigCodeBench Instruct | 46.1% | 45.8% |
| LMArena Coding | 1286 | 1292 |
| BigCodeBench Complete | 59% | 55.9% |
| Aider Polyglot | 28% | — |
| SciCode | 27.4% | — |
| LiveBench Coding | 51.4% | — |
| CadEval | 32% | — |
Agentic & Tool Use Claude 3.5 Haiku leads
Claude 3.5 Haiku: 28.0 (#95), Qwen2.5 72B Instruct: 22.1 (#133)
| Benchmark | Claude 3.5 Haiku | Qwen2.5 72B Instruct |
|---|---|---|
| BALROG | 19.3% | 16.2% |
| TheAgentCompany | — | 5.7% |
| METR Time Horizons | — | 35.8% |
Reasoning Qwen2.5 72B Instruct leads
Claude 3.5 Haiku: 17.7 (#290), Qwen2.5 72B Instruct: 22.3 (#199)
| Benchmark | Claude 3.5 Haiku | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Hard Prompts | 1251 | 1271 |
| DTBench | 56.7% | 62.9% |
| Epoch Capabilities Index | 127.15 | 129 |
| CritPt | 0% | — |
| LiveBench Reasoning | 28.1% | — |
| LiveBench Data Analysis | 48.5% | — |
| LMCA | — | 13.4% |
| BIG-Bench Hard | — | 79.8% |
| ForecastBench | — | 57.5 |
| HellaSwag | — | 84.8% |
| LiveBench | 43.5% | — |
| PIQA | — | 82.6% |
| WinoGrande | — | 82.3% |
Math Qwen2.5 72B Instruct leads
Claude 3.5 Haiku: 14.7 (#300), Qwen2.5 72B Instruct: 19.3 (#287)
| Benchmark | Claude 3.5 Haiku | Qwen2.5 72B Instruct |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 4.3% | 8.1% |
| Omni-MATH | 22.4% | 33% |
| LMArena Math | 1244 | 1283 |
| MATH Level 5 | 46.4% | 63.2% |
| LiveBench Math | 35.5% | — |
| FrontierMath (Feb 2025 set) | 0.3% | — |
Knowledge Qwen2.5 72B Instruct leads
Claude 3.5 Haiku: 18.7 (#281), Qwen2.5 72B Instruct: 27.0 (#253)
| Benchmark | Claude 3.5 Haiku | Qwen2.5 72B Instruct |
|---|---|---|
| GPQA Diamond | 38.1% | 49.1% |
| MMLU-Pro | 60.5% | 63.1% |
| Confabulations | 36.7% | 19.1% |
| GPQA (HELM) | 36.3% | 42.6% |
| LMArena Expert | 1208 | 1245 |
| MMLU | 74.3% | 85.3% |
| ARC (AI2) Challenge | — | 94.5% |
| TriviaQA | — | 71.9% |
Multimodal Not comparable
Claude 3.5 Haiku: 26.8 (#117), Qwen2.5 72B Instruct: —
| Benchmark | Claude 3.5 Haiku | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Vision | 1092 | — |
| GeoBench | 34% | — |
Multilingual Too close to call
Claude 3.5 Haiku: 40.0 (#218), Qwen2.5 72B Instruct: 41.0 (#213)
| Benchmark | Claude 3.5 Haiku | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Non-English | 1238 | 1252 |
| LMArena Chinese | 1229 | 1272 |
| LMArena French | 1264 | 1280 |
| LMArena German | 1237 | 1234 |
| LMArena Japanese | 1175 | 1180 |
| LMArena Korean | 1173 | 1188 |
| LMArena Russian | 1253 | 1264 |
| LMArena Spanish | 1261 | 1256 |
Instruction Following Qwen2.5 72B Instruct leads
Claude 3.5 Haiku: 62.9 (#234), Qwen2.5 72B Instruct: 65.5 (#221)
| Benchmark | Claude 3.5 Haiku | Qwen2.5 72B Instruct |
|---|---|---|
| IFEval | 79.2% | 80.6% |
| LMArena Instruction Following | 1241 | 1254 |
| LiveBench Instruction Following | 61.9% | — |
Long Context Too close to call
Claude 3.5 Haiku: 38.3 (#200), Qwen2.5 72B Instruct: 38.9 (#188)
| Benchmark | Claude 3.5 Haiku | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Longer Query | 1261 | 1282 |
Writing & Preference Qwen2.5 72B Instruct leads
Claude 3.5 Haiku: 42.7 (#234), Qwen2.5 72B Instruct: 46.7 (#215)
| Benchmark | Claude 3.5 Haiku | Qwen2.5 72B Instruct |
|---|---|---|
| LMArena Text | 1255 | 1269 |
| LMArena Creative Writing | 1233 | 1221 |
| WildBench | 76% | 80.2% |
| LMArena Multi-Turn | 1265 | 1272 |
| Short-Story Creative Writing | 73.5% | — |
| EQ-Bench Creative Writing | 1146 | — |
| LiveBench Language | 35.4% | — |
Frequently asked questions
Is Claude 3.5 Haiku better than Qwen2.5 72B Instruct?
Qwen2.5 72B Instruct is the stronger model overall, scoring 31.9 to 29.2 on the Noometry Index.
Is Claude 3.5 Haiku or Qwen2.5 72B Instruct better for coding?
They score almost the same on coding (32.9 vs 33.2); test both on your own repository before choosing.
How many benchmarks do Claude 3.5 Haiku and Qwen2.5 72B Instruct share?
33 benchmarks have published results for both models. Claude 3.5 Haiku has 49 scored results on Noometry and Qwen2.5 72B Instruct has 43.