Claude Fable 5 vs Kimi K3
Wins 7 of 8 areas
Coding · Agents · Reasoning · Facts · Images and charts · Math · Following instructions
Wins 1 of 8 areas
Long documents
Claude Fable 5 is the stronger all-rounder.Kimi K3 is cheaper and better at long documents.
Both rank among the ten best models we track in agents.
Scores updated · 51 tests both models report · How we compare
Where each one wins
Tests won in each of the eight areas we test. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
- ReasoningHard problems that need careful thinking41Claude Fable 54 of 6 tests · 1 tie
- MathCompetition and research-level math30Claude Fable 53 of 3 tests
- AgentsCarrying out multi-step tasks on its own32Claude Fable 53 of 7 tests · 2 ties
- CodingWriting and fixing software21Claude Fable 52 of 5 tests · 2 ties
- FactsGetting facts right instead of making them up21Claude Fable 52 of 3 tests
- Images and chartsUnderstanding pictures, charts and video10Claude Fable 51 of 1 test
- Following instructionsDoing exactly what it is asked10Claude Fable 51 of 1 test
- Long documentsFinding answers in very long texts01Kimi K31 of 1 test
Images and charts, long documents and following instructions rest on a single test each.
What it costs
Prices per million tokens, roughly 750,000 words. The bars show the cost of a million tokens read plus a million written.
Kimi K3 costs 70% less for the same work.
The biggest differences
The three tests each model wins by the widest margin. Scores are out of 100.
Where Claude Fable 5 pulls ahead
- Extremely hard research-level math problemsFrontierMath Tier 4+51.2points ahead
- Complex command-line tasks across many fieldsTerminal-Bench 4.0+29.8points ahead
- Abstract visual puzzles that people can solveARC-AGI-2+28.8points ahead
Where Kimi K3 pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+10.4points ahead
- Customer-service tasks in a simulated bankτ-Bench V3 · Banking+7.9points ahead
- Reasons across sets of long documentsAA-LCR+6.4points ahead
Every test, side by side
All 51 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingClaude Fable 5
- LiveBench · CodingClaude Fable 5 by 4.58681.5+4.5
- LMArena · WebDevKimi K3 by 31 rating points16241655+31 rating
- SciCodeClaude Fable 5 by 1.56159.5+1.5
- Terminal-Bench 2.1tie84.685tie
- LiveBench · Agentic Codingtie62.262.2tie
AgentsClaude Fable 5
- Terminal-Bench 4.0Claude Fable 5 by 29.842.412.6+29.8
- τ-Bench V3 · BankingKimi K3 by 7.938.146+7.9
- GDPValClaude Fable 5 by 3.955.651.7+3.9
- BrowseCompKimi K3 by 3.887.491.2+3.8
- AA ApexAgentsClaude Fable 5 by 243.341.3+2
- MCP Atlastie84.784.2tie
- OSWorld-Verifiedtie8584.8tie
ReasoningClaude Fable 5
- ARC-AGI-2Claude Fable 5 by 28.889.260.4+28.8
- SimpleBenchClaude Fable 5 by 21.281.960.7+21.2
- Humanity's Last ExamClaude Fable 5 by 8.655.546.9+8.6
- CritPtClaude Fable 5 by 5.228.623.4+5.2
- LiveBench · ReasoningKimi K3 by 189.790.7+1
- GPQA Diamondtie92.693.5tie
FactsClaude Fable 5
- SimpleQA VerifiedClaude Fable 5 by 20.170.750.6+20.1
- AA-Omniscience · AccuracyClaude Fable 5 by 17.865.347.6+17.8
- AA-Omniscience · Non-hallucinationKimi K3 by 10.436.446.8+10.4
Images and chartsClaude Fable 5
- MMMU-ProClaude Fable 5 by 3.784.280.5+3.7
MathClaude Fable 5
- FrontierMath Tier 4Claude Fable 5 by 51.290.239+51.2
- FrontierMath Tiers 1-3 (v2)Claude Fable 5 by 14.88772.2+14.8
- LiveBench · MathematicsClaude Fable 5 by 11.69684.4+11.6
Following instructionsClaude Fable 5
- LiveBench · Instruction FollowingClaude Fable 5 by 4.475.871.4+4.4
Other results24 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- AA-OmniscienceClaude Fable 5 by 23.643.319.7+23.6
- ToolathlonKimi K3 by 11.561.773.2+11.5
- Program BenchClaude Fable 5 by 8.586.377.8+8.5
- MCP Mark VerifiedKimi K3 by 7.187.494.5+7.1
- SWE-MarathonKimi K3 by 73542+7
- AA IntelligenceClaude Fable 5 by 649.643.6+6
- OfficeQA ProKimi K3 by 5.457.963.3+5.4
- livebench_languageClaude Fable 5 by 5.290.785.5+5.2
- PostTrainBenchClaude Fable 5 by 4.841.436.6+4.8
- JobBenchClaude Fable 5 by 4.557.452.9+4.5
- GDPval-AA v2Claude Fable 5 by 41 rating points17231682+41 rating
- ARC-AGI-1Claude Fable 5 by 498.594.5+4
- HLE (with tools)Claude Fable 5 by 463.859.8+4
- GDPval-AA v2 EloClaude Fable 5 by 37 rating points17231686+37 rating
- Agents' Last ExamKimi K3 by 2.625.728.3+2.6
- DeepSWEClaude Fable 5 by 2.57067.5+2.5
- livebench_data_analysisClaude Fable 5 by 1.880.578.7+1.8
- MLS Bench Litelower is betterKimi K3 by 1.649.948.3+1.6
- Toolathlon VerifiedClaude Fable 5 by 1.477.976.5+1.4
- DeepSWE 1.1Claude Fable 5 by 17069+1
- DeepSearchQA (F1)tie94.295tie
- AA Agentic Indextie5150.6tie
- Artificial Analysis Coding Indextie76.576.2tie
- SpreadsheetBench 2tie34.734.8tie
Questions people ask
Which is better, Claude Fable 5 or Kimi K3?
Claude Fable 5 wins seven of the eight areas we test: coding, agents, reasoning, facts, images and charts, math and following instructions. Kimi K3 wins long documents, and costs 70% less.
Which is better for coding?
Claude Fable 5. It wins 2 of the 5 coding tests both models report; Kimi K3 wins 1, and 2 are ties.
Which is cheaper?
Claude Fable 5 costs $10.00 per million input tokens and $50.00 per million output tokens; Kimi K3 costs $3.00 and $15.00. That makes Kimi K3 about 70% cheaper for the same work.
How do you compare the two?
We use the 51 benchmark tests both models have published scores on. The verdict counts the 27 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 24 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.