Model comparison
Claude Fable 5 vs Kimi K3
Claude Fable 5 is the stronger model overall, scoring 66.8 to 59.5 on the Noometry Index. Kimi K3 costs 3.3× less per token, which makes it the better buy when Claude Fable 5's lead doesn't matter for your workload.
Last verified . 51 shared benchmarks.
Summary
- They share 51 benchmarks with published results for both. Claude Fable 5 scores higher in 8 categories and Kimi K3 in 2 categories; 6 gaps are clear of the uncertainty.
- The widest gap is in math, where Claude Fable 5 leads 88.5 to 74.2.
- The biggest single-benchmark swing is FrontierMath Tier 4: 90.2% for Claude Fable 5 and 39% for Kimi K3.
- Kimi K3 is cheaper at $3 / $15 per million input/output tokens, against $10 / $50 for Claude Fable 5.
- Kimi K3 accepts more context: 1.05M tokens versus 1M.
- Kimi K3 has downloadable open weights; the other is API-only.
Side by side
| Claude Fable 5 | Kimi K3 | |
|---|---|---|
| Provider | Anthropic | Moonshot AI |
| Noometry Index | 66.8 | 59.5 |
| Released | 2026-06-07 | 2026-07-16 |
| Weights | Proprietary | Open |
| Context window | 1M | 1.05M |
| Max output | 128K | 1.05M |
| Input $ / M tokens | $10 | $3 |
| Output $ / M tokens | $50 | $15 |
| Results tracked | 62 | 53 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Claude Fable 5 leads
Claude Fable 5: 70.6 (#4), Kimi K3: 61.0 (#10)
| Benchmark | Claude Fable 5 | Kimi K3 |
|---|---|---|
| DeepSWE | 69.9% | 68.5% |
| FrontierCode | 53.5% | 44.2% |
| LMArena WebDev | 1625 | 1654 |
| FrontierSWE | 47% | 25.9% |
| SciCode | 61% | 59.5% |
| WeirdML | 91.9% | 82.6% |
| LMArena Coding | 1519 | 1508 |
| ALE-Bench | 2,041 | 1,524 |
| GSO | 78.4% | — |
| MirrorCode | 63.9% | — |
Agentic & Tool Use Claude Fable 5 leads
Claude Fable 5: 54.0 (#2), Kimi K3: 41.8 (#20)
| Benchmark | Claude Fable 5 | Kimi K3 |
|---|---|---|
| APEX-Agents | 63.6% | 50.6% |
| τ²-bench Banking | 39.7% | 37.1% |
| PostTrainBench | 41.8% | 32% |
| GBAEval | 74.5% | 48.3% |
| GDP.pdf | 30% | 19% |
| Vending-Bench 2 | 5,680 | 5,165 |
| Remote Labor Index | 16.1% | — |
| LMArena Search | 1230 | — |
Reasoning Claude Fable 5 leads
Claude Fable 5: 76.8 (#6), Kimi K3: 63.0 (#17)
| Benchmark | Claude Fable 5 | Kimi K3 |
|---|---|---|
| ARC-AGI-2 | 89.2% | 60.4% |
| SimpleBench | 81.9% | 60.7% |
| NYT Connections (extended) | 92.7% | 93.6% |
| ARC-AGI-1 | 98.5% | 94.5% |
| CritPt | 28.6% | 23.4% |
| Chess Puzzles | 41% | 39% |
| LMArena Hard Prompts | 1508 | 1496 |
| Mystery Game Puzzles | 52% | 26% |
| DTBench | 98.4% | 91.2% |
| LMCA | 61.1% | 52.7% |
| Surface Evolver Bench | 95% | 95% |
| Epoch Capabilities Index | 162.06 | 157.45 |
| Kagi LLM Benchmark | 91.4% | — |
| EnigmaEval | 39.3% | — |
| EBR-Bench | 39.5% | — |
| Bench to the Future 3 | 0.13 | — |
| ForecastBench | — | 61.1 |
Math Claude Fable 5 leads
Claude Fable 5: 88.5 (#5), Kimi K3: 74.2 (#16)
| Benchmark | Claude Fable 5 | Kimi K3 |
|---|---|---|
| FrontierMath (Tiers 1-3) | 87% | 72.2% |
| FrontierMath Tier 4 | 90.2% | 39% |
| OTIS Mock AIME 2024-2025 | 100% | 97.2% |
| ProofBench | 95% | 87% |
| LMArena Math | 1519 | 1491 |
| MathArena Final-Answer Competitions | — | 87.8% |
| FrontierMath Erdős | 0% | — |
Knowledge Too close to call
Claude Fable 5: 62.2 (#25), Kimi K3: 63.2 (#21)
| Benchmark | Claude Fable 5 | Kimi K3 |
|---|---|---|
| GPQA Diamond | 85.9% | 93.1% |
| SimpleQA Verified | 70.7% | 50.6% |
| LMArena Expert | 1534 | 1521 |
Multimodal Claude Fable 5 leads
Claude Fable 5: 45.3 (#17), Kimi K3: 37.8 (#70)
| Benchmark | Claude Fable 5 | Kimi K3 |
|---|---|---|
| Blueprint-Bench 2 | 38.6% | 29.5% |
| Furniture Assembly | 35.8% | 34.2% |
| LMArena Vision | 1324 | — |
| LMArena Document | 1496 | — |
Multilingual Claude Fable 5 leads
Claude Fable 5: 57.3 (#9), Kimi K3: 56.3 (#21)
| Benchmark | Claude Fable 5 | Kimi K3 |
|---|---|---|
| LMArena Non-English | 1481 | 1466 |
| LMArena Chinese | 1543 | 1529 |
| LMArena French | 1505 | 1491 |
| LMArena German | 1486 | 1488 |
| LMArena Japanese | 1506 | 1487 |
| LMArena Korean | 1488 | 1458 |
| LMArena Russian | 1504 | 1482 |
| LMArena Spanish | 1498 | 1472 |
Instruction Following Too close to call
Claude Fable 5: 78.6 (#8), Kimi K3: 77.7 (#14)
| Benchmark | Claude Fable 5 | Kimi K3 |
|---|---|---|
| LMArena Instruction Following | 1502 | 1483 |
Long Context Too close to call
Claude Fable 5: 46.3 (#23), Kimi K3: 45.8 (#29)
| Benchmark | Claude Fable 5 | Kimi K3 |
|---|---|---|
| LMArena Longer Query | 1509 | 1494 |
Writing & Preference Too close to call
Claude Fable 5: 75.9 (#5), Kimi K3: 76.6 (#4)
| Benchmark | Claude Fable 5 | Kimi K3 |
|---|---|---|
| LMArena Text | 1491 | 1476 |
| LMArena Creative Writing | 1494 | 1454 |
| EQ-Bench Creative Writing | 1943 | 2082 |
| EQ-Bench 4 | 1340 | 1339 |
| LMArena Multi-Turn | 1504 | 1488 |
Frequently asked questions
Is Claude Fable 5 better than Kimi K3?
Claude Fable 5 is the stronger model overall, scoring 66.8 to 59.5 on the Noometry Index. Kimi K3 costs 3.3× less per token, which makes it the better buy when Claude Fable 5's lead doesn't matter for your workload.
Which is cheaper, Claude Fable 5 or Kimi K3?
Kimi K3 is cheaper. It lists at $3 per million input tokens and $15 per million output tokens; Claude Fable 5 lists at $10 and $50.
Is Claude Fable 5 or Kimi K3 better for coding?
Claude Fable 5 scores higher on coding benchmarks: 70.6 versus 61.0 in the Noometry coding category.
Which has the bigger context window?
Kimi K3 does, with 1.05M tokens against 1M.
How many benchmarks do Claude Fable 5 and Kimi K3 share?
51 benchmarks have published results for both models. Claude Fable 5 has 62 scored results on Noometry and Kimi K3 has 53.