Model comparison
Claude Haiku 4.5 vs Qwen3.8 Max
Qwen3.8 Max is the stronger model overall, scoring 56.8 to 39.5 on the Noometry Index. Claude Haiku 4.5 costs 1.5× less per token, which makes it the better buy when Qwen3.8 Max's lead doesn't matter for your workload.
Last verified . 28 shared benchmarks.
Alibaba (Qwen)Summary
- They share 28 benchmarks with published results for both. Claude Haiku 4.5 scores higher in 0 categories and Qwen3.8 Max in 10 categories; 10 gaps are clear of the uncertainty.
- The widest gap is in reasoning, where Qwen3.8 Max leads 54.4 to 15.1.
- The biggest single-benchmark swing is NYT Connections (extended): 14.3% for Claude Haiku 4.5 and 88.3% for Qwen3.8 Max.
- Claude Haiku 4.5 is cheaper at $1 / $5 per million input/output tokens, against $2 / $6 for Qwen3.8 Max.
- Qwen3.8 Max accepts more context: 1M tokens versus 200K.
Side by side
| Claude Haiku 4.5 | Qwen3.8 Max | |
|---|---|---|
| Provider | Anthropic | Alibaba (Qwen) |
| Noometry Index | 39.5 | 56.8 |
| Released | 2025-10-15 | 2026-08-02 |
| Weights | Proprietary | Proprietary |
| Context window | 200K | 1M |
| Max output | 64K | 131K |
| Input $ / M tokens | $1 | $2 |
| Output $ / M tokens | $5 | $6 |
| Results tracked | 53 | 39 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Qwen3.8 Max leads
Claude Haiku 4.5: 44.0 (#78), Qwen3.8 Max: 53.5 (#29)
| Benchmark | Claude Haiku 4.5 | Qwen3.8 Max |
|---|---|---|
| LMArena WebDev | 1330 | 1674 |
| SciCode | 43.3% | 53.2% |
| LMArena Coding | 1453 | 1502 |
| DeepSWE | — | 57.5% |
| SWE-bench Verified (bash only) | 66.6% | — |
| SWE-bench Multilingual | 64.7% | — |
| FrontierSWE | — | 17.8% |
| WeirdML | 45.4% | — |
| ALE-Bench | 653.48 | — |
Agentic & Tool Use Qwen3.8 Max leads
Claude Haiku 4.5: 33.6 (#52), Qwen3.8 Max: 45.4 (#14)
| Benchmark | Claude Haiku 4.5 | Qwen3.8 Max |
|---|---|---|
| Terminal-Bench | 35.5% | — |
| APEX-Agents | — | 63.3% |
| Berkeley Function Calling Leaderboard | 68.7% | — |
| τ²-bench Banking | — | 55.1% |
| DeepResearch Bench | 45.5% | — |
| BALROG | 31.2% | — |
| ExploitBench | 13.7% | — |
| GDP.pdf | — | 23.2% |
| Vending-Bench 2 | 458.89 | — |
Reasoning Qwen3.8 Max leads
Claude Haiku 4.5: 15.1 (#320), Qwen3.8 Max: 54.4 (#26)
| Benchmark | Claude Haiku 4.5 | Qwen3.8 Max |
|---|---|---|
| NYT Connections (extended) | 14.3% | 88.3% |
| CritPt | 0% | 20% |
| Chess Puzzles | 8% | 40% |
| LMArena Hard Prompts | 1420 | 1496 |
| DTBench | 73.6% | 92% |
| LMCA | 30.9% | 46.2% |
| Epoch Capabilities Index | 142.41 | 156.41 |
| ARC-AGI-2 | 4% | — |
| ARC-AGI-1 | 47.7% | — |
| Mystery Game Puzzles | — | 38% |
| ForecastBench | 61.4 | — |
Math Qwen3.8 Max leads
Claude Haiku 4.5: 44.9 (#78), Qwen3.8 Max: 73.2 (#20)
| Benchmark | Claude Haiku 4.5 | Qwen3.8 Max |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 66.7% | 100% |
| LMArena Math | 1396 | 1499 |
| FrontierMath (Tiers 1-3) | — | 74.7% |
| FrontierMath Tier 4 | — | 46.3% |
| ProofBench | — | 58% |
| Omni-MATH | 56.1% | — |
| MATH Level 5 | 96.4% | — |
| FrontierMath (Feb 2025 set) | 5.9% | — |
| FrontierMath Tier 4 (v1) | 2.1% | — |
Knowledge Qwen3.8 Max leads
Claude Haiku 4.5: 37.7 (#153), Qwen3.8 Max: 61.7 (#27)
| Benchmark | Claude Haiku 4.5 | Qwen3.8 Max |
|---|---|---|
| GPQA Diamond | 71.2% | 92.7% |
| SimpleQA Verified | 13.2% | 47.3% |
| LMArena Expert | 1442 | 1507 |
| MMLU-Pro | 77.7% | — |
| Vectara Hallucination Rate | 9.8% | — |
| GPQA (HELM) | 60.5% | — |
Multimodal Qwen3.8 Max leads
Claude Haiku 4.5: 26.8 (#118), Qwen3.8 Max: 37.2 (#75)
| Benchmark | Claude Haiku 4.5 | Qwen3.8 Max |
|---|---|---|
| LMArena Vision | — | 1314 |
| Blueprint-Bench 2 | 0% | — |
| Furniture Assembly | — | 20% |
| LMArena Document | 1420 | — |
Multilingual Qwen3.8 Max leads
Claude Haiku 4.5: 49.9 (#129), Qwen3.8 Max: 56.7 (#18)
| Benchmark | Claude Haiku 4.5 | Qwen3.8 Max |
|---|---|---|
| LMArena Non-English | 1377 | 1472 |
| LMArena Chinese | 1417 | 1538 |
| LMArena French | 1408 | 1503 |
| LMArena German | 1375 | 1483 |
| LMArena Japanese | 1339 | 1467 |
| LMArena Korean | 1347 | 1461 |
| LMArena Russian | 1381 | 1481 |
| LMArena Spanish | 1420 | 1492 |
Instruction Following Qwen3.8 Max leads
Claude Haiku 4.5: 71.4 (#149), Qwen3.8 Max: 77.6 (#17)
| Benchmark | Claude Haiku 4.5 | Qwen3.8 Max |
|---|---|---|
| LMArena Instruction Following | 1414 | 1479 |
| IFEval | 80.1% | — |
Long Context Qwen3.8 Max leads
Claude Haiku 4.5: 43.6 (#92), Qwen3.8 Max: 45.6 (#31)
| Benchmark | Claude Haiku 4.5 | Qwen3.8 Max |
|---|---|---|
| LMArena Longer Query | 1427 | 1489 |
Writing & Preference Qwen3.8 Max leads
Claude Haiku 4.5: 57.9 (#123), Qwen3.8 Max: 67.1 (#30)
| Benchmark | Claude Haiku 4.5 | Qwen3.8 Max |
|---|---|---|
| LMArena Text | 1396 | 1483 |
| LMArena Creative Writing | 1372 | 1479 |
| LMArena Multi-Turn | 1409 | 1489 |
| WildBench | 83.9% | — |
| EQ-Bench 4 | 1064 | — |
Frequently asked questions
Is Claude Haiku 4.5 better than Qwen3.8 Max?
Qwen3.8 Max is the stronger model overall, scoring 56.8 to 39.5 on the Noometry Index. Claude Haiku 4.5 costs 1.5× less per token, which makes it the better buy when Qwen3.8 Max's lead doesn't matter for your workload.
Which is cheaper, Claude Haiku 4.5 or Qwen3.8 Max?
Claude Haiku 4.5 is cheaper. It lists at $1 per million input tokens and $5 per million output tokens; Qwen3.8 Max lists at $2 and $6.
Is Claude Haiku 4.5 or Qwen3.8 Max better for coding?
Qwen3.8 Max scores higher on coding benchmarks: 53.5 versus 44.0 in the Noometry coding category.
Which has the bigger context window?
Qwen3.8 Max does, with 1M tokens against 200K.
How many benchmarks do Claude Haiku 4.5 and Qwen3.8 Max share?
28 benchmarks have published results for both models. Claude Haiku 4.5 has 53 scored results on Noometry and Qwen3.8 Max has 39.