Claude Fable 5 vs Qwen3 235B A22B Thinking 2507
Claude Fable 5 leads Qwen3 235B A22B Thinking 2507 on the Artificial Analysis Intelligence Index, 49.6 to 12.7, and Qwen3 235B A22B Thinking 2507 costs 14.3x less per output token ($3.50 vs $50.00 per million). Updated October 22, 2026; 44 benchmarks have a result for both.
vs boards won of 44 shared
Claude Fable 5 or Qwen3 235B A22B Thinking 2507: which to pick
Only Claude Fable 5 has a coding score, 79.3 on the Vibe Coding Index; Qwen3 235B A22B Thinking 2507 has none, so the two cannot be ranked on coding.
Pick Qwen3 235B A22B Thinking 2507 for output tokens 14.3x cheaper ($3.50 vs $50.00 per million).
Model AClaude Fable 5Anthropic79.3/1006 independent benchmarks$2.16 per solved taskModel BQwen3 235B A22B Thinking 2507QwenNot benchmarkedNo coding benchmarkNo cost per solved task
- Shared results
- 44
- Claude Fable 5 only
- 91
- Qwen3 235B A22B Thinking 2507 only
- 2
- Like-for-like categories
- 8 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 Fable 5 has a coding score, so no call is made.
One side not measured
Intelligence
AA Intelligence Index
Claude Fable 5
Claude Fable 5 leads, 49.6 against 12.7.
Same board, both measured
Agentic work
AA Agentic Index
Claude Fable 5
Claude Fable 5 leads, 50.7 against 1.3.
Same board, both measured
Reasoning
GPQA Diamond
No call
Only Claude Fable 5 has a GPQA Diamond result, 93.2%.
Directional only
Knowledge
MMLU Pro
No call
Only Claude Fable 5 has a MMLU Pro result, 91.5%.
Directional only
Web design
Design Arena: websites
Claude Fable 5
Claude Fable 5 leads, 1295 against 1052.
Same board, both measured
Long documents
Listed context window
Claude Fable 5
Claude Fable 5 takes 1M tokens against 128K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Qwen3 235B A22B Thinking 2507
Claude Fable 5 $0.035, Qwen3 235B A22B Thinking 2507 $0.0022 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
No call
Only Claude Fable 5 has the number, $2.16.
Directional only
At a glance
- Coding index
- Claude Fable 5 79.3 from 6 benchmarks, Qwen3 235B A22B Thinking 2507 not benchmarked.
- Widest category
- Agentic work, 42.5 points apart on average over 2 shared benchmarks.
- Widest single gap
- AA Coding Index: Claude Fable 5 76.5 against Qwen3 235B A22B Thinking 2507 22.1, 54.4 points apart on a 0 to 100 scale.
- Output price
- Claude Fable 5 $50.00 against Qwen3 235B A22B Thinking 2507 $3.50 per million output tokens, 14.3x apart.
- Context window
- Claude Fable 5 takes 1M tokens, Qwen3 235B A22B Thinking 2507 128K.
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 Fable 5 | Bars | Qwen3 235B A22B Thinking 2507 | Ties |
|---|---|---|---|---|
| Coding | 4 | 0 | ||
| Intelligence | 1 | 0 | ||
| Agentic work | 2 | 0 | ||
| Reasoning | 5 | 0 | ||
| Math | 3 | 0 | ||
| Knowledge | 5 | 0 | ||
| Web design | 6 | 0 | ||
| General | 18 | 0 | ||
| All shared boards | 44 | 0 |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index40
| Benchmark | Claude Fable 5 | Qwen3 235B A22B Thinking 2507 | Leader |
|---|---|---|---|
| LiveBench coding | 98.3% | Not measured | Only Claude Fable 5 measured |
| LMArena WebDev | 1624 | Not measured | Only Claude Fable 5 measured |
| AA Coding Index | 76.5 | 22.1 | Claude Fable 5 |
| LiveCodeBench (Vals AI) | 89.8% | Not measured | Only Claude Fable 5 measured |
| SWE-bench Verified (Vals AI) | 95.0% | Not measured | Only Claude Fable 5 measured |
| Terminal-Bench 4 (Vals AI) | 41.4% | Not measured | Only Claude Fable 5 measured |
| Vibe Code Bench (Vals AI) | 90.3% | Not measured | Only Claude Fable 5 measured |
| Code migration (Vals AI) | 55.1% | Not measured | Only Claude Fable 5 measured |
| ProgramBench (Vals AI) | 2.0% | Not measured | Only Claude Fable 5 measured |
| Terminal-Bench 2.1 (Vals AI) | 80.5% | Not measured | Only Claude Fable 5 measured |
| SciCode | 61.0% | Not measured | Only Claude Fable 5 measured |
| Terminal-Bench Hard (Artificial Analysis run) | 62.9% | Not measured | Only Claude Fable 5 measured |
| FrontierCode | 53.5% | Not measured | Only Claude Fable 5 measured |
| FrontierSWE | 47.0% | Not measured | Only Claude Fable 5 measured |
| GSO-Bench | 78.4% | Not measured | Only Claude Fable 5 measured |
| WeirdML | 91.9% | 41.0% | Claude Fable 5 |
| MirrorCode | 63.9% | Not measured | Only Claude Fable 5 measured |
| ALE-Bench | 2041.3 | Not measured | Only Claude Fable 5 measured |
| LMArena Text: coding | 1518 | 1424 | Claude Fable 5 |
| LMArena Text: software and IT | 1513 | 1431 | Claude Fable 5 |
| LMArena WebDev: React | 1631 | Not measured | Only Claude Fable 5 measured |
| LMArena WebDev: HTML | 1651 | Not measured | Only Claude Fable 5 measured |
| LMArena WebDev: from an image | 1623 | Not measured | Only Claude Fable 5 measured |
| SWE Atlas: Codebase QnA | 39.0% | Not measured | Only Claude Fable 5 measured |
| SWE Atlas: Refactoring | 54.8% | Not measured | Only Claude Fable 5 measured |
| SWE Atlas: Test Writing | 55.6% | Not measured | Only Claude Fable 5 measured |
| Neither model has a result on the other 17 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | Claude Fable 5 | Qwen3 235B A22B Thinking 2507 | Leader |
|---|---|---|---|
| AA Intelligence Index | 49.6 | 12.7 | Claude Fable 5 |
| Epoch Capabilities Index | 163.4 | Not measured | Only Claude Fable 5 measured |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index20
| Benchmark | Claude Fable 5 | Qwen3 235B A22B Thinking 2507 | Leader |
|---|---|---|---|
| AA Agentic Index | 50.7 | 1.3 | Claude Fable 5 |
| Design Arena: agents building web apps | 1261 | Not measured | Only Claude Fable 5 measured |
| Design Arena: agents building full-stack apps | 1280 | Not measured | Only Claude Fable 5 measured |
| Design Arena: agents building mobile apps | 1263 | Not measured | Only Claude Fable 5 measured |
| Design Arena: agents building Android apps | 1283 | Not measured | Only Claude Fable 5 measured |
| Design Arena: agents building Godot games | 1292 | Not measured | Only Claude Fable 5 measured |
| Design Arena: agents building games | 1285 | Not measured | Only Claude Fable 5 measured |
| Design Arena: HTML slides | 1216 | Not measured | Only Claude Fable 5 measured |
| Design Arena: PowerPoint slides | 1269 | Not measured | Only Claude Fable 5 measured |
| LiveBench research | 86.9% | Not measured | Only Claude Fable 5 measured |
| LiveBench browser | 92.0% | Not measured | Only Claude Fable 5 measured |
| LiveBench long agentic | 53.2% | Not measured | Only Claude Fable 5 measured |
| Finance Agent (Vals AI) | 56.3% | Not measured | Only Claude Fable 5 measured |
| Tax Agent Bench (Vals AI) | 38.3% | Not measured | Only Claude Fable 5 measured |
| Legal Research Bench (Vals AI) | 49.5% | Not measured | Only Claude Fable 5 measured |
| Excel Modeling Benchmark (Vals AI) | 73.7% | Not measured | Only Claude Fable 5 measured |
| Terminal-Bench-Science (Vals AI) | 15.7% | Not measured | Only Claude Fable 5 measured |
| τ²-Bench Telecom | 98.5% | Not measured | Only Claude Fable 5 measured |
| APEX-Agents | 63.6% | Not measured | Only Claude Fable 5 measured |
| Remote Labor Index | 16.1% | Not measured | Only Claude Fable 5 measured |
| PostTrainBench | 41.8% | Not measured | Only Claude Fable 5 measured |
| Vending-Bench 2 | $5,680 | -$11 | Claude Fable 5 |
| LMArena Search | 1229 | Not measured | Only Claude Fable 5 measured |
| MCP Atlas | 83.3% | Not measured | Only Claude Fable 5 measured |
| Neither model has a result on the other 18 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index50
| Benchmark | Claude Fable 5 | Qwen3 235B A22B Thinking 2507 | Leader |
|---|---|---|---|
| GPQA Diamond | 93.2% | Not measured | Only Claude Fable 5 measured |
| LiveBench reasoning | 89.7% | Not measured | Only Claude Fable 5 measured |
| MMMU Pro (Vals AI) | 89.3% | Not measured | Only Claude Fable 5 measured |
| GPQA Diamond (Artificial Analysis run) | 92.6% | Not measured | Only Claude Fable 5 measured |
| Humanity's Last Exam (Artificial Analysis run) | 55.5% | Not measured | Only Claude Fable 5 measured |
| GPQA Diamond (Epoch AI run) | 85.9% | 80.1% | Claude Fable 5 |
| ARC-AGI | 98.5% | Not measured | Only Claude Fable 5 measured |
| ARC-AGI-2 | 89.2% | Not measured | Only Claude Fable 5 measured |
| SimpleBench | 81.9% | Not measured | Only Claude Fable 5 measured |
| Chess puzzles | 47.0% | Not measured | Only Claude Fable 5 measured |
| Mystery game puzzles | 52.0% | 9.0% | Claude Fable 5 |
| EBR-bench | 39.5% | Not measured | Only Claude Fable 5 measured |
| LMArena Text: hard prompts | 1507 | 1416 | Claude Fable 5 |
| LMArena Text: expert | 1534 | 1463 | Claude Fable 5 |
| LMArena Text: hard prompts in English | 1509 | 1432 | Claude Fable 5 |
| LMArena Vision: diagrams | 1357 | Not measured | Only Claude Fable 5 measured |
| LMArena Vision: homework | 1361 | Not measured | Only Claude Fable 5 measured |
| Kagi LLM Benchmark | 91.4% | Not measured | Only Claude Fable 5 measured |
| Neither model has a result on the other 5 boards here | |||
MathLike-for-likePublic boards, shown as published, kept out of the index30
| Benchmark | Claude Fable 5 | Qwen3 235B A22B Thinking 2507 | Leader |
|---|---|---|---|
| OTIS Mock AIME | 100.0% | 86.7% | Claude Fable 5 |
| FrontierMath, tiers 1 to 3 | 87.0% | Not measured | Only Claude Fable 5 measured |
| FrontierMath, tier 4 | 90.2% | Not measured | Only Claude Fable 5 measured |
| ProofBench | 95.0% | Not measured | Only Claude Fable 5 measured |
| LMArena Text: math | 1516 | 1412 | Claude Fable 5 |
| LMArena Text: mathematical work | 1515 | 1435 | Claude Fable 5 |
| MathArena | 51.5% | Not measured | Only Claude Fable 5 measured |
| Neither model has a result on the other 7 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index50
| Benchmark | Claude Fable 5 | Qwen3 235B A22B Thinking 2507 | Leader |
|---|---|---|---|
| MMLU Pro | 91.5% | Not measured | Only Claude Fable 5 measured |
| LegalBench (Vals AI) | 88.6% | Not measured | Only Claude Fable 5 measured |
| CorpFin (Vals AI) | 71.8% | Not measured | Only Claude Fable 5 measured |
| TaxEval (Vals AI) | 76.9% | Not measured | Only Claude Fable 5 measured |
| MortgageTax (Vals AI) | 68.9% | Not measured | Only Claude Fable 5 measured |
| MedCode (Vals AI) | 56.1% | Not measured | Only Claude Fable 5 measured |
| SimpleQA Verified | 70.7% | 40.4% | Claude Fable 5 |
| LMArena Text: science | 1506 | 1433 | Claude Fable 5 |
| LMArena Text: medicine and healthcare | 1480 | 1456 | Claude Fable 5 |
| LMArena Text: legal and government | 1498 | 1415 | Claude Fable 5 |
| LMArena Text: business and finance | 1480 | 1413 | Claude Fable 5 |
| PRBench Finance | 53.9% | Not measured | Only Claude Fable 5 measured |
| PRBench Legal | 52.6% | Not measured | Only Claude Fable 5 measured |
| Neither model has a result on the other 7 boards here | |||
Web designLike-for-likePublic boards, shown as published, kept out of the index60
| Benchmark | Claude Fable 5 | Qwen3 235B A22B Thinking 2507 | Leader |
|---|---|---|---|
| Design Arena: websites | 1295 | 1052 | Claude Fable 5 |
| Design Arena: UI components | 1316 | 942 | Claude Fable 5 |
| Design Arena: game dev | 1335 | 956 | Claude Fable 5 |
| Design Arena: data viz | 1305 | 950 | Claude Fable 5 |
| Design Arena: 3D | 1318 | 997 | Claude Fable 5 |
| Design Arena: all code categories | 1310 | 1032 | Claude Fable 5 |
| Design Arena: SVG | 1301 | Not measured | Only Claude Fable 5 measured |
| Design Arena: ASCII art | 1329 | Not measured | Only Claude Fable 5 measured |
| LiveBench design | 84.7% | Not measured | Only Claude Fable 5 measured |
GeneralLike-for-likePublic boards, shown as published, kept out of the index180
| Benchmark | Claude Fable 5 | Qwen3 235B A22B Thinking 2507 | Leader |
|---|---|---|---|
| LiveBench data analysis | 80.5% | Not measured | Only Claude Fable 5 measured |
| Public Benefits Bench (Vals AI) | 70.4% | Not measured | Only Claude Fable 5 measured |
| SAGE (Vals AI) | 51.9% | Not measured | Only Claude Fable 5 measured |
| MedScribe (Vals AI) | 88.5% | Not measured | Only Claude Fable 5 measured |
| Vals Multimodal Index | 74.2% | Not measured | Only Claude Fable 5 measured |
| IFBench | 63.5% | Not measured | Only Claude Fable 5 measured |
| Long context reasoning (AA-LCR) | 82.3% | Not measured | Only Claude Fable 5 measured |
| DTBench | 98.4% | 80.3% | Claude Fable 5 |
| LMCA | 61.2% | 29.3% | Claude Fable 5 |
| Fiction.LiveBench | Not measured | 75.0% | Only Qwen3 235B A22B Thinking 2507 measured |
| Creative writing (Lech Mazur) | Not measured | 82.4% | Only Qwen3 235B A22B Thinking 2507 measured |
| Surface Evolver Bench | 95.0% | Not measured | Only Claude Fable 5 measured |
| Furniture assembly | 35.8% | Not measured | Only Claude Fable 5 measured |
| LMArena Text | 1491 | 1415 | Claude Fable 5 |
| LMArena Text: instruction following | 1502 | 1386 | Claude Fable 5 |
| LMArena Text: creative writing | 1494 | 1384 | Claude Fable 5 |
| LMArena Vision | 1325 | Not measured | Only Claude Fable 5 measured |
| LMArena Text: English | 1498 | 1429 | Claude Fable 5 |
| LMArena Text: non-English | 1482 | 1397 | Claude Fable 5 |
| LMArena Text: Chinese | 1543 | 1481 | Claude Fable 5 |
| LMArena Text: German | 1486 | 1408 | Claude Fable 5 |
| LMArena Text: Japanese | 1504 | 1386 | Claude Fable 5 |
| LMArena Text: Korean | 1487 | 1375 | Claude Fable 5 |
| LMArena Text: French | 1505 | Not measured | Only Claude Fable 5 measured |
| LMArena Text: Spanish | 1498 | 1389 | Claude Fable 5 |
| LMArena Text: Russian | 1503 | 1401 | Claude Fable 5 |
| LMArena Text: Polish | 1494 | 1390 | Claude Fable 5 |
| LMArena Text: multi-turn | 1504 | 1406 | Claude Fable 5 |
| LMArena Text: longer queries | 1510 | 1400 | Claude Fable 5 |
| LMArena Text: writing and language | 1501 | 1388 | Claude Fable 5 |
| LMArena Text: entertainment and media | 1482 | 1370 | Claude Fable 5 |
| LMArena Vision: English | 1323 | Not measured | Only Claude Fable 5 measured |
| LMArena Vision: Chinese | 1373 | Not measured | Only Claude Fable 5 measured |
| LMArena Vision: reading text | 1337 | Not measured | Only Claude Fable 5 measured |
| LMArena Vision: humor | 1341 | Not measured | Only Claude Fable 5 measured |
| LMArena Vision: creative writing | 1338 | Not measured | Only Claude Fable 5 measured |
| LMArena Document | 1496 | Not measured | Only Claude Fable 5 measured |
| EQ-Bench 4 | 1340 | Not measured | Only Claude Fable 5 measured |
| Neither model has a result on the other 10 boards here | |||
Image generationNot comparablePublic boards, shown as published, kept out of the index0 boards, none shared
| Benchmark | Claude Fable 5 | Qwen3 235B A22B Thinking 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
Claude Fable 5$0.035Fits in one request
Qwen3 235B A22B Thinking 2507$0.0022Fits in one request
Qwen3 235B A22B Thinking 2507 costs less
Repository review
50K input + 3K output tokens
Claude Fable 5$0.650Fits in one request
Qwen3 235B A22B Thinking 2507$0.033Fits in one request
Qwen3 235B A22B Thinking 2507 costs less
Agent loop
200K input + 10K output tokens
Claude Fable 5$2.50Fits in one request
Qwen3 235B A22B Thinking 2507$0.125Exceeds the context window
Qwen3 235B A22B Thinking 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 Fable 5 | Qwen3 235B A22B Thinking 2507 |
|---|---|---|
| Maker | Anthropic | Qwen |
| Weights | No public weights listed | Open weights |
| Released | June 9, 2026 | July 25, 2025 |
| Context window | 1M | 128K |
| Input $/Mtok | $10.00 | $0.45 |
| Output $/Mtok | $50.00 | $3.50 |
| Coding index | 79.3 | Not benchmarked |
| Independent measurements behind it | 6 | 0 |
| Cost per solved task | $2.16 | Not computable |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Claude Fable 5 better than Qwen3 235B A22B Thinking 2507 for coding?
Only Claude Fable 5 has a coding score, 79.3 on the Vibe Coding Index; Qwen3 235B A22B Thinking 2507 has none, so the two cannot be ranked on coding. Pick Qwen3 235B A22B Thinking 2507 for output tokens 14.3x cheaper ($3.50 vs $50.00 per million).
Which is cheaper, Claude Fable 5 or Qwen3 235B A22B Thinking 2507?
Claude Fable 5 costs $10.00 per million input tokens and $50.00 per million output tokens; Qwen3 235B A22B Thinking 2507 costs $0.45 and $3.50. Cost per solved task is known for Claude Fable 5 only.
Which has the bigger context window, Claude Fable 5 or Qwen3 235B A22B Thinking 2507?
Claude Fable 5, with 1M tokens against 128K.
How many benchmarks is the Claude Fable 5 vs Qwen3 235B A22B Thinking 2507 comparison based on?
44 benchmarks have a result for both models as of October 22, 2026. Claude Fable 5's coding index rests on 6 independent benchmarks and Qwen3 235B A22B Thinking 2507's on 0. Where only one model has a result, the page says so instead of naming a winner.