Claude Haiku 4.5 vs Qwen3 235B A22B Instruct 2507
Claude Haiku 4.5 leads Qwen3 235B A22B Instruct 2507 on the Artificial Analysis Intelligence Index, 16.9 to 12.0, and Qwen3 235B A22B Instruct 2507 costs 6.5x less per output token ($0.775 vs $5.00 per million). Updated October 22, 2026; 13 benchmarks have a result for both.
Claude Haiku 4.5 or Qwen3 235B A22B Instruct 2507: which to pick
Only Claude Haiku 4.5 has a coding score, 35.6 on the Vibe Coding Index; Qwen3 235B A22B Instruct 2507 has none, so the two cannot be ranked on coding.
Pick Qwen3 235B A22B Instruct 2507 for output tokens 6.5x cheaper ($0.775 vs $5.00 per million) and a larger context window (262K against 200K tokens).
- Shared results
- 13
- Claude Haiku 4.5 only
- 85
- Qwen3 235B A22B Instruct 2507 only
- 11
- Like-for-like categories
- 6 of 9
Only benchmarks that tested both models count toward a pick; a result only one model has is shown but not compared.
Which one for your work
Each card is decided by one number both models carry. Where only one has it, the card says so and makes no call.
Coding
Vibe Coding Index, six public coding benchmarks
No call
Only Claude Haiku 4.5 has a coding score, so no call is made.
One side not measured
Intelligence
AA Intelligence Index
Claude Haiku 4.5
Claude Haiku 4.5 leads, 16.9 against 12.0.
Same board, both measured
Agentic work
AA Agentic Index
No call
Only Claude Haiku 4.5 has a AA Agentic Index result, 8.0.
Directional only
Reasoning
GPQA Diamond
No call
Only Claude Haiku 4.5 has a GPQA Diamond result, 72.2%.
Directional only
Math
AIME
No call
Only Claude Haiku 4.5 has a AIME result, 82.7%.
Directional only
Knowledge
MMLU Pro
No call
Only Claude Haiku 4.5 has a MMLU Pro result, 78.7%.
Directional only
Web design
Design Arena: websites
Claude Haiku 4.5
Claude Haiku 4.5 leads, 1122 against 1058.
Same board, both measured
Long documents
Listed context window
Qwen3 235B A22B Instruct 2507
Qwen3 235B A22B Instruct 2507 takes 262K tokens against 200K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Qwen3 235B A22B Instruct 2507
Claude Haiku 4.5 $0.0035, Qwen3 235B A22B Instruct 2507 $0.0005 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
No call
Only Claude Haiku 4.5 has the number, $0.56.
Directional only
At a glance
- Coding index
- Claude Haiku 4.5 35.6 from 4 benchmarks, Qwen3 235B A22B Instruct 2507 not benchmarked.
- Widest category
- Reasoning, 19.7 points apart on average over 2 shared benchmarks.
- Widest single gap
- ARC-AGI: Claude Haiku 4.5 47.7% against Qwen3 235B A22B Instruct 2507 11.0%, 36.7 points apart on a 0 to 100 scale.
- Output price
- Claude Haiku 4.5 $5.00 against Qwen3 235B A22B Instruct 2507 $0.775 per million output tokens, 6.5x apart.
- Context window
- Claude Haiku 4.5 takes 200K tokens, Qwen3 235B A22B Instruct 2507 262K.
Benchmark by benchmark
First the score: boards each model leads, per category, where both have a result. Then every board, a category to a fold; bars are each result on a 0 to 100 scale, so compare the two in a row, not one row with another.
| Category | Claude Haiku 4.5 | Bars | Qwen3 235B A22B Instruct 2507 | Ties |
|---|---|---|---|---|
| Coding | 1 | 0 | ||
| Intelligence | 1 | 0 | ||
| Agentic work | 1 | 0 | ||
| Reasoning | 2 | 0 | ||
| Web design | 6 | 0 | ||
| General | 1 | 1 | ||
| All shared boards | 12 | 1 |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index10
| Benchmark | Claude Haiku 4.5 | Qwen3 235B A22B Instruct 2507 | Leader |
|---|---|---|---|
| SWE-bench Verified | 66.6% | Not measured | Only Claude Haiku 4.5 measured |
| SWE-bench Multilingual | 64.7% | Not measured | Only Claude Haiku 4.5 measured |
| LiveBench coding | 48.1% | Not measured | Only Claude Haiku 4.5 measured |
| LMArena WebDev | 1330 | Not measured | Only Claude Haiku 4.5 measured |
| AA Coding Index | 43.9 | Not measured | Only Claude Haiku 4.5 measured |
| LiveCodeBench (Vals AI) | 41.2% | Not measured | Only Claude Haiku 4.5 measured |
| SWE-bench Verified (Vals AI) | 66.6% | Not measured | Only Claude Haiku 4.5 measured |
| Vibe Code Bench (Vals AI) | 11.4% | Not measured | Only Claude Haiku 4.5 measured |
| Code migration (Vals AI) | 7.5% | Not measured | Only Claude Haiku 4.5 measured |
| ProgramBench (Vals AI) | 0.0% | Not measured | Only Claude Haiku 4.5 measured |
| Terminal-Bench 2.1 (Vals AI) | 43.8% | Not measured | Only Claude Haiku 4.5 measured |
| LiveCodeBench (Artificial Analysis run) | Not measured | 78.8% | Only Qwen3 235B A22B Instruct 2507 measured |
| SciCode | Not measured | 41.4% | Only Qwen3 235B A22B Instruct 2507 measured |
| Terminal-Bench Hard (Artificial Analysis run) | Not measured | 15.2% | Only Qwen3 235B A22B Instruct 2507 measured |
| Terminal-Bench | 35.5% | Not measured | Only Claude Haiku 4.5 measured |
| WeirdML | 45.4% | 38.7% | Claude Haiku 4.5 |
| LMArena Text: coding | 1454 | Not measured | Only Claude Haiku 4.5 measured |
| LMArena Text: software and IT | 1439 | Not measured | Only Claude Haiku 4.5 measured |
| LMArena WebDev: React | 1326 | Not measured | Only Claude Haiku 4.5 measured |
| LMArena WebDev: HTML | 1318 | Not measured | Only Claude Haiku 4.5 measured |
| Neither model has a result on the other 23 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | Claude Haiku 4.5 | Qwen3 235B A22B Instruct 2507 | Leader |
|---|---|---|---|
| AA Intelligence Index | 16.9 | 12.0 | Claude Haiku 4.5 |
| Epoch Capabilities Index | 142.2 | Not measured | Only Claude Haiku 4.5 measured |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | Claude Haiku 4.5 | Qwen3 235B A22B Instruct 2507 | Leader |
|---|---|---|---|
| AA Agentic Index | 8.0 | Not measured | Only Claude Haiku 4.5 measured |
| LiveBench research | 42.2% | Not measured | Only Claude Haiku 4.5 measured |
| LiveBench browser | 40.0% | Not measured | Only Claude Haiku 4.5 measured |
| LiveBench long agentic | 11.1% | Not measured | Only Claude Haiku 4.5 measured |
| Finance Agent (Vals AI) | 31.0% | Not measured | Only Claude Haiku 4.5 measured |
| Tax Agent Bench (Vals AI) | 5.9% | Not measured | Only Claude Haiku 4.5 measured |
| Legal Research Bench (Vals AI) | 10.6% | Not measured | Only Claude Haiku 4.5 measured |
| Excel Modeling Benchmark (Vals AI) | 22.7% | Not measured | Only Claude Haiku 4.5 measured |
| τ²-Bench Telecom | Not measured | 53.2% | Only Qwen3 235B A22B Instruct 2507 measured |
| BALROG | 31.2% | Not measured | Only Claude Haiku 4.5 measured |
| DeepResearch Bench | 45.5% | Not measured | Only Claude Haiku 4.5 measured |
| Vending-Bench 2 | $459 | Not measured | Only Claude Haiku 4.5 measured |
| Berkeley Function Calling Leaderboard (BFCL) | 68.7% | 52.1% | Claude Haiku 4.5 |
| MCP Atlas | 40.2% | Not measured | Only Claude Haiku 4.5 measured |
| Neither model has a result on the other 28 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index20
| Benchmark | Claude Haiku 4.5 | Qwen3 235B A22B Instruct 2507 | Leader |
|---|---|---|---|
| GPQA Diamond | 72.2% | Not measured | Only Claude Haiku 4.5 measured |
| LiveBench reasoning | 61.7% | Not measured | Only Claude Haiku 4.5 measured |
| MMMU Pro (Vals AI) | 46.1% | Not measured | Only Claude Haiku 4.5 measured |
| GPQA Diamond (Artificial Analysis run) | Not measured | 79.0% | Only Qwen3 235B A22B Instruct 2507 measured |
| Humanity's Last Exam (Artificial Analysis run) | Not measured | 15.9% | Only Qwen3 235B A22B Instruct 2507 measured |
| GPQA Diamond (Epoch AI run) | 71.2% | Not measured | Only Claude Haiku 4.5 measured |
| ARC-AGI | 47.7% | 11.0% | Claude Haiku 4.5 |
| ARC-AGI-2 | 4.0% | 1.3% | Claude Haiku 4.5 |
| Chess puzzles | 8.0% | Not measured | Only Claude Haiku 4.5 measured |
| LMArena Text: hard prompts | 1421 | Not measured | Only Claude Haiku 4.5 measured |
| LMArena Text: expert | 1443 | Not measured | Only Claude Haiku 4.5 measured |
| LMArena Text: hard prompts in English | 1435 | Not measured | Only Claude Haiku 4.5 measured |
| Neither model has a result on the other 11 boards here | |||
MathDirectional onlyPublic boards, shown as published, kept out of the index8 boards, none shared
| Benchmark | Claude Haiku 4.5 | Qwen3 235B A22B Instruct 2507 | Leader |
|---|---|---|---|
| AIME | 82.7% | Not measured | Only Claude Haiku 4.5 measured |
| MGSM (Vals AI) | 92.2% | Not measured | Only Claude Haiku 4.5 measured |
| AA Math Index | Not measured | 91.0 | Only Qwen3 235B A22B Instruct 2507 measured |
| AIME 2025 (Artificial Analysis run) | Not measured | 91.0% | Only Qwen3 235B A22B Instruct 2507 measured |
| OTIS Mock AIME | 66.7% | Not measured | Only Claude Haiku 4.5 measured |
| MATH level 5 | 96.4% | Not measured | Only Claude Haiku 4.5 measured |
| LMArena Text: math | 1398 | Not measured | Only Claude Haiku 4.5 measured |
| LMArena Text: mathematical work | 1430 | Not measured | Only Claude Haiku 4.5 measured |
| Neither model has a result on the other 6 boards here | |||
KnowledgeDirectional onlyPublic boards, shown as published, kept out of the index14 boards, none shared
| Benchmark | Claude Haiku 4.5 | Qwen3 235B A22B Instruct 2507 | Leader |
|---|---|---|---|
| MMLU Pro | 78.7% | Not measured | Only Claude Haiku 4.5 measured |
| LegalBench (Vals AI) | 81.2% | Not measured | Only Claude Haiku 4.5 measured |
| CaseLaw (Vals AI) | 56.5% | Not measured | Only Claude Haiku 4.5 measured |
| CorpFin (Vals AI) | 60.6% | Not measured | Only Claude Haiku 4.5 measured |
| TaxEval (Vals AI) | 67.5% | Not measured | Only Claude Haiku 4.5 measured |
| MortgageTax (Vals AI) | 62.2% | Not measured | Only Claude Haiku 4.5 measured |
| MedQA (Vals AI) | 79.6% | Not measured | Only Claude Haiku 4.5 measured |
| MedCode (Vals AI) | 32.7% | Not measured | Only Claude Haiku 4.5 measured |
| MMLU Pro (Artificial Analysis run) | Not measured | 84.3% | Only Qwen3 235B A22B Instruct 2507 measured |
| SimpleQA Verified | 13.2% | Not measured | Only Claude Haiku 4.5 measured |
| LMArena Text: science | 1406 | Not measured | Only Claude Haiku 4.5 measured |
| LMArena Text: medicine and healthcare | 1394 | Not measured | Only Claude Haiku 4.5 measured |
| LMArena Text: legal and government | 1400 | Not measured | Only Claude Haiku 4.5 measured |
| LMArena Text: business and finance | 1400 | Not measured | Only Claude Haiku 4.5 measured |
| Neither model has a result on the other 6 boards here | |||
Web designLike-for-likePublic boards, shown as published, kept out of the index60
| Benchmark | Claude Haiku 4.5 | Qwen3 235B A22B Instruct 2507 | Leader |
|---|---|---|---|
| Design Arena: websites | 1122 | 1058 | Claude Haiku 4.5 |
| Design Arena: UI components | 1097 | 963 | Claude Haiku 4.5 |
| Design Arena: game dev | 1096 | 951 | Claude Haiku 4.5 |
| Design Arena: data viz | 1124 | 1070 | Claude Haiku 4.5 |
| Design Arena: 3D | 1071 | 994 | Claude Haiku 4.5 |
| Design Arena: all code categories | 1114 | 1038 | Claude Haiku 4.5 |
| Design Arena: SVG | 1025 | Not measured | Only Claude Haiku 4.5 measured |
| Design Arena: ASCII art | 1140 | Not measured | Only Claude Haiku 4.5 measured |
| LiveBench design | 59.9% | Not measured | Only Claude Haiku 4.5 measured |
GeneralLike-for-likePublic boards, shown as published, kept out of the index11
| Benchmark | Claude Haiku 4.5 | Qwen3 235B A22B Instruct 2507 | Leader |
|---|---|---|---|
| LiveBench data analysis | 59.3% | Not measured | Only Claude Haiku 4.5 measured |
| Public Benefits Bench (Vals AI) | 54.3% | Not measured | Only Claude Haiku 4.5 measured |
| SAGE (Vals AI) | 31.8% | Not measured | Only Claude Haiku 4.5 measured |
| MedScribe (Vals AI) | 85.2% | Not measured | Only Claude Haiku 4.5 measured |
| Vals Multimodal Index | 42.9% | Not measured | Only Claude Haiku 4.5 measured |
| IFBench | Not measured | 51.2% | Only Qwen3 235B A22B Instruct 2507 measured |
| Long context reasoning (AA-LCR) | Not measured | 72.0% | Only Qwen3 235B A22B Instruct 2507 measured |
| DTBench | 73.6% | 78.4% | Qwen3 235B A22B Instruct 2507 |
| LMCA | 30.9% | 23.5% | Claude Haiku 4.5 |
| ExploitBench | 13.7% | Not measured | Only Claude Haiku 4.5 measured |
| LMArena Text | 1397 | Not measured | Only Claude Haiku 4.5 measured |
| LMArena Text: instruction following | 1414 | Not measured | Only Claude Haiku 4.5 measured |
| LMArena Text: creative writing | 1371 | Not measured | Only Claude Haiku 4.5 measured |
| LMArena Text: English | 1414 | Not measured | Only Claude Haiku 4.5 measured |
| LMArena Text: non-English | 1377 | Not measured | Only Claude Haiku 4.5 measured |
| LMArena Text: Chinese | 1418 | Not measured | Only Claude Haiku 4.5 measured |
| LMArena Text: German | 1374 | Not measured | Only Claude Haiku 4.5 measured |
| LMArena Text: Japanese | 1338 | Not measured | Only Claude Haiku 4.5 measured |
| LMArena Text: Korean | 1346 | Not measured | Only Claude Haiku 4.5 measured |
| LMArena Text: French | 1411 | Not measured | Only Claude Haiku 4.5 measured |
| LMArena Text: Spanish | 1421 | Not measured | Only Claude Haiku 4.5 measured |
| LMArena Text: Russian | 1381 | Not measured | Only Claude Haiku 4.5 measured |
| LMArena Text: Polish | 1343 | Not measured | Only Claude Haiku 4.5 measured |
| LMArena Text: multi-turn | 1409 | Not measured | Only Claude Haiku 4.5 measured |
| LMArena Text: longer queries | 1427 | Not measured | Only Claude Haiku 4.5 measured |
| LMArena Text: writing and language | 1386 | Not measured | Only Claude Haiku 4.5 measured |
| LMArena Text: entertainment and media | 1369 | Not measured | Only Claude Haiku 4.5 measured |
| LMArena Document | 1420 | Not measured | Only Claude Haiku 4.5 measured |
| MultiChallenge | 50.5% | Not measured | Only Claude Haiku 4.5 measured |
| EQ-Bench 4 | 1064 | Not measured | Only Claude Haiku 4.5 measured |
| Neither model has a result on the other 18 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Claude Haiku 4.5 | Qwen3 235B A22B Instruct 2507 | Leader |
|---|---|---|---|
| Neither model has a result on 5 boards here | |||
What each workload costs
Three fixed token mixes at each model's listed price, so per-token rates become a bill.
Chat turn
1K input + 500 output tokens
Qwen3 235B A22B Instruct 2507 costs less
Repository review
50K input + 3K output tokens
Qwen3 235B A22B Instruct 2507 costs less
Agent loop
200K input + 10K output tokens
Qwen3 235B A22B Instruct 2507 costs less
Listed OpenRouter rates, without cache or batch discounts. A workload that does not fit a model's context window would need splitting.
Specification differences
The leading value in a row is marked; a row with a missing value names no leader. Our two rows are marked in red.
| Specification | Anthropic Claude Haiku 4.5 | Qwen3 235B A22B Instruct 2507 |
|---|---|---|
| Maker | Anthropic | Qwen |
| Weights | No public weights listed | Open weights |
| Released | October 15, 2025 | July 21, 2025 |
| Context window | 200K | 262K |
| Input $/Mtok | $1.00 | $0.145 |
| Output $/Mtok | $5.00 | $0.775 |
| Coding index | 35.6 | Not benchmarked |
| Independent measurements behind it | 4 | 0 |
| Cost per solved task | $0.56 | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Claude Haiku 4.5 better than Qwen3 235B A22B Instruct 2507 for coding?
Only Claude Haiku 4.5 has a coding score, 35.6 on the Vibe Coding Index; Qwen3 235B A22B Instruct 2507 has none, so the two cannot be ranked on coding. Pick Qwen3 235B A22B Instruct 2507 for output tokens 6.5x cheaper ($0.775 vs $5.00 per million) and a larger context window (262K against 200K tokens).
Which is cheaper, Claude Haiku 4.5 or Qwen3 235B A22B Instruct 2507?
Claude Haiku 4.5 costs $1.00 per million input tokens and $5.00 per million output tokens; Qwen3 235B A22B Instruct 2507 costs $0.145 and $0.775. Cost per solved task is known for Claude Haiku 4.5 only.
Which has the bigger context window, Claude Haiku 4.5 or Qwen3 235B A22B Instruct 2507?
Qwen3 235B A22B Instruct 2507, with 262K tokens against 200K.
How many benchmarks is the Claude Haiku 4.5 vs Qwen3 235B A22B Instruct 2507 comparison based on?
13 benchmarks have a result for both models as of October 22, 2026. Claude Haiku 4.5's coding index rests on 4 independent benchmarks and Qwen3 235B A22B Instruct 2507's on 0. Where only one model has a result, the page says so instead of naming a winner.