Up to four models
Muse Spark 1.3 vs Kimi K3
On the index (GPT-4 = 100), Muse Spark 1.3 scores 169 and Kimi K3 168. Side by side on the same third-party tests, with price and context. The link in your address bar shares this exact comparison.
169153–186
WeightsClosed
ReleasedSep 2026
Price in / out$1.25 / $4.25
Context1.05M
Core tests taken4 of 9
Kimi K3
168152–183
WeightsOpen
ReleasedJul 2026
Price in / out$0.19 / $16
Context1.05M
Core tests taken6 of 9
Muse Spark 1.3 and Kimi K3 are too close to call (2 points apart, within the uncertainty), ahead on 3 of 4 shared core tests, and costs 2.1× less.
Test by testthird-party results
Every result as a table
| Test | Muse Spark 1.3 | Kimi K3 |
|---|---|---|
| Reasoning & knowledge | ||
| LiveBench Data Analysis | 79.6% | 78.7% |
| LiveBench IF | 78.0% | 71.4% |
| LiveBench Language | 82.8% | 85.5% |
| LiveBench Reasoning | 89.6% | 90.7% |
| SimpleBench | 81.8% | 60.7% |
| Maths | ||
| FrontierMath (tiers 1-3) | 74.4% | 72.2% |
| AIME-style maths (OTIS mock) | 99.2% | 97.2% |
| LiveBench Mathematics | 96.0% | 84.4% |
| ProofBench v1.1 | 55.0% | 87.0% |
| Chess puzzles | 37.0% | 37.0% |
| FrontierMath (tier 4) | 46.3% | 39.0% |
| Coding | ||
| Terminal-Bench 4.0 | 10.6% | 17.2% |
| LiveBench Agentic Coding | 64.1% | 62.2% |
| LiveBench Coding | 81.1% | 81.5% |
| Code Migration | 27.6% | 16.1% |
| CyberBench v1.1 | 69.4% | 75.2% |
| IOI | 43.9% | 48.9% |
| ProgramBench | 0.5% | 2.0% |
| Vibe Code Bench v1.1 | 82.9% | 85.0% |
| FrontierSWE | 25.8% | 25.9% |
| MirrorCode | 54.4% | 54.4% |
| Agents | ||
| APEX-Agents | 58.6% | 47.8% |
| Terminal-Bench-Science | 4.3% | 1.4% |
| DTBench | 96.5% | 91.2% |
| LMCA | 53.9% | 52.7% |
| Mystery game puzzles | 25.0% | 26.0% |
| Professional work | ||
| Finance Agent (v2) | 58.9% | 53.1% |
| EMB | 62.7% | 66.7% |
| Legal Research Bench | 40.9% | 46.2% |
| Tax Agent Bench | 37.9% | 33.4% |
Tests at least two of these models have taken. Bold rows are the core tests.