Claude Fable 5 vs Kimi K2 Thinking
Claude Fable 5 leads Kimi K2 Thinking on the Artificial Analysis Intelligence Index, 49.6 to 22.0, and Kimi K2 Thinking costs 20.0x less per output token ($2.50 vs $50.00 per million). Updated October 22, 2026; 18 benchmarks have a result for both.
vs boards won of 18 shared
Claude Fable 5 or Kimi K2 Thinking: which to pick
Only Claude Fable 5 has a coding score, 79.3 on the Vibe Coding Index; Kimi K2 Thinking has none, so the two cannot be ranked on coding.
Pick Kimi K2 Thinking for output tokens 20.0x cheaper ($2.50 vs $50.00 per million) and a lower cost per solved task ($0.61 against $2.16).
Model AClaude Fable 5Anthropic79.3/1006 independent benchmarks$2.16 per solved taskModel BKimi K2 ThinkingMoonshot AINot benchmarked1 independent benchmark$0.61 per solved task
- Shared results
- 18
- Claude Fable 5 only
- 109
- Kimi K2 Thinking only
- 12
- Like-for-like categories
- 7 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 22.0.
Same board, both measured
Agentic work
AA Agentic Index
No call
Only Claude Fable 5 has a AA Agentic Index result, 50.7.
Directional only
Reasoning
GPQA Diamond
Claude Fable 5
Claude Fable 5 leads, 93.2% against 78.5%.
Same board, both measured
Math
AIME
No call
Only Kimi K2 Thinking has a AIME result, 85.4%.
Directional only
Knowledge
MMLU Pro
Claude Fable 5
Claude Fable 5 leads, 91.5% against 81.1%.
Same board, both measured
Web design
Design Arena: websites
Claude Fable 5
Claude Fable 5 leads, 1295 against 1113.
Same board, both measured
Long documents
Listed context window
Claude Fable 5
Claude Fable 5 takes 1M tokens against 262K.
Listed by OpenRouter
Everyday cost
1K input + 500 output tokens
Kimi K2 Thinking
Claude Fable 5 $0.035, Kimi K2 Thinking $0.0019 per turn.
List prices
Cost per solved task
One SWE-bench Verified task, solved
Kimi K2 Thinking
Claude Fable 5 $2.16, Kimi K2 Thinking $0.61.
Published attempt cost and pass rate
At a glance
- Coding index
- Claude Fable 5 79.3 from 6 benchmarks, Kimi K2 Thinking not benchmarked.
- Widest category
- Coding, 38.1 points apart on average over 5 shared benchmarks.
- Widest single gap
- WeirdML: Claude Fable 5 91.9% against Kimi K2 Thinking 42.8%, 49.1 points apart on a 0 to 100 scale.
- Output price
- Claude Fable 5 $50.00 against Kimi K2 Thinking $2.50 per million output tokens, 20.0x apart.
- Context window
- Claude Fable 5 takes 1M tokens, Kimi K2 Thinking 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 Fable 5 | Bars | Kimi K2 Thinking | Ties |
|---|---|---|---|---|
| Coding | 5 | 0 | ||
| Intelligence | 2 | 0 | ||
| Agentic work | 1 | 0 | ||
| Reasoning | 3 | 0 | ||
| Knowledge | 4 | 0 | ||
| Web design | 1 | 0 | ||
| General | 1 | 1 | ||
| All shared boards | 17 | 1 |
CodingLike-for-likeSix of these boards feed the Vibe Coding Index50
| Benchmark | Claude Fable 5 | Kimi K2 Thinking | Leader |
|---|---|---|---|
| SWE-bench Verified | Not measured | 63.4% | Only Kimi K2 Thinking measured |
| 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 | Not measured | Only Claude Fable 5 measured |
| LiveCodeBench (Vals AI) | 89.8% | 63.1% | Claude Fable 5 |
| SWE-bench Verified (Vals AI) | 95.0% | 60.2% | Claude Fable 5 |
| 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 |
| LiveCodeBench (Artificial Analysis run) | Not measured | 85.3% | Only Kimi K2 Thinking measured |
| SciCode | 61.0% | Not measured | Only Claude Fable 5 measured |
| Terminal-Bench Hard (Artificial Analysis run) | 62.9% | 31.1% | Claude Fable 5 |
| Terminal-Bench | Not measured | 35.7% | Only Kimi K2 Thinking 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% | 42.8% | Claude Fable 5 |
| MirrorCode | 63.9% | Not measured | Only Claude Fable 5 measured |
| ALE-Bench | 2041.3 | 597.5 | Claude Fable 5 |
| LMArena Text: coding | 1518 | Not measured | Only Claude Fable 5 measured |
| LMArena Text: software and IT | 1513 | Not measured | Only Claude Fable 5 measured |
| 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 |
| Neither model has a result on the other 13 boards here | |||
IntelligenceLike-for-likePublic boards, shown as published, kept out of the index20
| Benchmark | Claude Fable 5 | Kimi K2 Thinking | Leader |
|---|---|---|---|
| AA Intelligence Index | 49.6 | 22.0 | Claude Fable 5 |
| Epoch Capabilities Index | 163.4 | 145.8 | Claude Fable 5 |
Agentic workLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | Claude Fable 5 | Kimi K2 Thinking | Leader |
|---|---|---|---|
| AA Agentic Index | 50.7 | Not measured | Only Claude Fable 5 measured |
| 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 | 1264 | Not measured | Only Claude Fable 5 measured |
| Design Arena: agents building Android apps | 1284 | 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 | 1227 | Not measured | Only Claude Fable 5 measured |
| Design Arena: PowerPoint slides | 1270 | 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% | 93.0% | Claude Fable 5 |
| APEX-Agents | 63.6% | Not measured | Only Claude Fable 5 measured |
| METR time horizons | Not measured | 59.2% | Only Kimi K2 Thinking 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 | Not measured | Only Claude Fable 5 measured |
| LMArena Search | 1229 | Not measured | Only Claude Fable 5 measured |
| Neither model has a result on the other 17 boards here | |||
ReasoningLike-for-likePublic boards, shown as published, kept out of the index30
| Benchmark | Claude Fable 5 | Kimi K2 Thinking | Leader |
|---|---|---|---|
| GPQA Diamond | 93.2% | 78.5% | Claude Fable 5 |
| 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% | 83.8% | Claude Fable 5 |
| Humanity's Last Exam (Artificial Analysis run) | 55.5% | 23.8% | Claude Fable 5 |
| GPQA Diamond (Epoch AI run) | 85.9% | Not measured | Only Claude Fable 5 measured |
| 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% | Not measured | Only Claude Fable 5 measured |
| EBR-bench | 39.5% | Not measured | Only Claude Fable 5 measured |
| LMArena Text: hard prompts | 1507 | Not measured | Only Claude Fable 5 measured |
| LMArena Text: expert | 1534 | Not measured | Only Claude Fable 5 measured |
| LMArena Text: hard prompts in English | 1509 | Not measured | Only Claude Fable 5 measured |
| 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 4 boards here | |||
MathDirectional onlyPublic boards, shown as published, kept out of the index10 boards, none shared
| Benchmark | Claude Fable 5 | Kimi K2 Thinking | Leader |
|---|---|---|---|
| AIME | Not measured | 85.4% | Only Kimi K2 Thinking measured |
| MGSM (Vals AI) | Not measured | 90.2% | Only Kimi K2 Thinking measured |
| AA Math Index | Not measured | 94.7 | Only Kimi K2 Thinking measured |
| AIME 2025 (Artificial Analysis run) | Not measured | 94.7% | Only Kimi K2 Thinking measured |
| OTIS Mock AIME | 100.0% | Not measured | Only Claude Fable 5 measured |
| 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 | Not measured | Only Claude Fable 5 measured |
| LMArena Text: mathematical work | 1515 | Not measured | Only Claude Fable 5 measured |
| Neither model has a result on the other 3 boards here | |||
KnowledgeLike-for-likePublic boards, shown as published, kept out of the index40
| Benchmark | Claude Fable 5 | Kimi K2 Thinking | Leader |
|---|---|---|---|
| MMLU Pro | 91.5% | 81.1% | Claude Fable 5 |
| LegalBench (Vals AI) | 88.6% | 80.2% | Claude Fable 5 |
| CaseLaw (Vals AI) | Not measured | 65.7% | Only Kimi K2 Thinking measured |
| CorpFin (Vals AI) | 71.8% | 60.6% | Claude Fable 5 |
| TaxEval (Vals AI) | 76.9% | 71.7% | Claude Fable 5 |
| MortgageTax (Vals AI) | 68.9% | Not measured | Only Claude Fable 5 measured |
| MedQA (Vals AI) | Not measured | 92.6% | Only Kimi K2 Thinking measured |
| MedCode (Vals AI) | 56.1% | Not measured | Only Claude Fable 5 measured |
| MMLU Pro (Artificial Analysis run) | Not measured | 84.8% | Only Kimi K2 Thinking measured |
| SimpleQA Verified | 70.7% | Not measured | Only Claude Fable 5 measured |
| LMArena Text: science | 1506 | Not measured | Only Claude Fable 5 measured |
| LMArena Text: medicine and healthcare | 1480 | Not measured | Only Claude Fable 5 measured |
| LMArena Text: legal and government | 1498 | Not measured | Only Claude Fable 5 measured |
| LMArena Text: business and finance | 1480 | Not measured | Only Claude Fable 5 measured |
| Neither model has a result on the other 4 boards here | |||
Web designLike-for-likePublic boards, shown as published, kept out of the index10
| Benchmark | Claude Fable 5 | Kimi K2 Thinking | Leader |
|---|---|---|---|
| Design Arena: websites | 1295 | 1113 | Claude Fable 5 |
| Design Arena: UI components | 1316 | Not measured | Only Claude Fable 5 measured |
| Design Arena: game dev | 1335 | Not measured | Only Claude Fable 5 measured |
| Design Arena: data viz | 1304 | Not measured | Only Claude Fable 5 measured |
| Design Arena: 3D | 1318 | Not measured | Only Claude Fable 5 measured |
| Design Arena: all code categories | 1310 | Not measured | Only Claude Fable 5 measured |
| Design Arena: SVG | 1302 | 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 index11
| Benchmark | Claude Fable 5 | Kimi K2 Thinking | 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% | 68.1% | Kimi K2 Thinking |
| Long context reasoning (AA-LCR) | 82.3% | 72.0% | Claude Fable 5 |
| DTBench | 98.4% | Not measured | Only Claude Fable 5 measured |
| LMCA | 61.2% | Not measured | Only Claude Fable 5 measured |
| CL-bench | Not measured | 17.6% | Only Kimi K2 Thinking 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 | Not measured | Only Claude Fable 5 measured |
| LMArena Text: instruction following | 1502 | Not measured | Only Claude Fable 5 measured |
| LMArena Text: creative writing | 1494 | Not measured | Only Claude Fable 5 measured |
| LMArena Vision | 1325 | Not measured | Only Claude Fable 5 measured |
| LMArena Text: English | 1498 | Not measured | Only Claude Fable 5 measured |
| LMArena Text: non-English | 1482 | Not measured | Only Claude Fable 5 measured |
| LMArena Text: Chinese | 1543 | Not measured | Only Claude Fable 5 measured |
| LMArena Text: German | 1486 | Not measured | Only Claude Fable 5 measured |
| LMArena Text: Japanese | 1504 | Not measured | Only Claude Fable 5 measured |
| LMArena Text: Korean | 1487 | Not measured | Only Claude Fable 5 measured |
| LMArena Text: French | 1505 | Not measured | Only Claude Fable 5 measured |
| LMArena Text: Spanish | 1498 | Not measured | Only Claude Fable 5 measured |
| LMArena Text: Russian | 1503 | Not measured | Only Claude Fable 5 measured |
| LMArena Text: Polish | 1494 | Not measured | Only Claude Fable 5 measured |
| LMArena Text: multi-turn | 1504 | Not measured | Only Claude Fable 5 measured |
| LMArena Text: longer queries | 1510 | Not measured | Only Claude Fable 5 measured |
| LMArena Text: writing and language | 1501 | Not measured | Only Claude Fable 5 measured |
| LMArena Text: entertainment and media | 1482 | Not measured | Only Claude Fable 5 measured |
| 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 |
| 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 | Kimi K2 Thinking | 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
Kimi K2 Thinking$0.0019Fits in one request
Kimi K2 Thinking costs less
Repository review
50K input + 3K output tokens
Claude Fable 5$0.650Fits in one request
Kimi K2 Thinking$0.037Fits in one request
Kimi K2 Thinking costs less
Agent loop
200K input + 10K output tokens
Claude Fable 5$2.50Fits in one request
Kimi K2 Thinking$0.145Fits in one request
Kimi K2 Thinking 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 | MoonshotAI Kimi K2 Thinking |
|---|---|---|
| Maker | Anthropic | Moonshot AI |
| Weights | No public weights listed | Open weights |
| Released | June 9, 2026 | November 6, 2025 |
| Context window | 1M | 262K |
| Input $/Mtok | $10.00 | $0.60 |
| Output $/Mtok | $50.00 | $2.50 |
| Coding index | 79.3 | Not benchmarked |
| Independent measurements behind it | 6 | 1 |
| Cost per solved task | $2.16 | $0.61 |
| Source | OpenRouter listing | OpenRouter listing |
Questions
Is Claude Fable 5 better than Kimi K2 Thinking for coding?
Only Claude Fable 5 has a coding score, 79.3 on the Vibe Coding Index; Kimi K2 Thinking has none, so the two cannot be ranked on coding. Pick Kimi K2 Thinking for output tokens 20.0x cheaper ($2.50 vs $50.00 per million) and a lower cost per solved task ($0.61 against $2.16).
Which is cheaper, Claude Fable 5 or Kimi K2 Thinking?
Claude Fable 5 costs $10.00 per million input tokens and $50.00 per million output tokens; Kimi K2 Thinking costs $0.60 and $2.50. Per solved SWE-bench Verified task, Claude Fable 5 costs $2.16 and Kimi K2 Thinking $0.61.
Which has the bigger context window, Claude Fable 5 or Kimi K2 Thinking?
Claude Fable 5, with 1M tokens against 262K.
How many benchmarks is the Claude Fable 5 vs Kimi K2 Thinking comparison based on?
18 benchmarks have a result for both models as of October 22, 2026. Claude Fable 5's coding index rests on 6 independent benchmarks and Kimi K2 Thinking's on 1. Where only one model has a result, the page says so instead of naming a winner.