Up to four models
Claude Haiku 5.5 vs Muse Spark 1.3
On the index (GPT-4 = 100), Claude Haiku 5.5 scores 174 and Muse Spark 1.3 172. Side by side on the same third-party tests, with price and context. The link in your address bar shares this exact comparison.
Claude Haiku 5.5
174158–191
WeightsClosed
ReleasedOct 2026
Price in / out$0.10 / $0.50
Context1M
Core tests taken7 of 9
Muse Spark 1.3172156–188
WeightsClosed
ReleasedSep 2026
Price in / out$1.25 / $4.25
Context1.05M
Core tests taken4 of 9
Claude Haiku 5.5 and Muse Spark 1.3 are too close to call (2 points apart, within the uncertainty), ahead on 2 of 4 shared core tests, and costs 10× less.
Test by testthird-party results
Claude Haiku 5.5Muse Spark 1.3
Muse Spark 1.3Every result as a table
| Test | Claude Haiku 5.5 | Muse Spark 1.3 |
|---|---|---|
| Reasoning & knowledge | ||
| LiveBench Data Analysis | 72.8% | 79.6% |
| LiveBench IF | 66.5% | 78.0% |
| LiveBench Language | 63.5% | 82.8% |
| LiveBench Reasoning | 84.6% | 89.6% |
| Maths | ||
| FrontierMath (tiers 1-3) | 75.1% | 74.4% |
| AIME-style maths (OTIS mock) | 98.9% | 99.2% |
| LiveBench Mathematics | 92.8% | 96.0% |
| ProofBench v1.1 | 89.0% | 55.0% |
| Chess puzzles | 55.0% | 37.0% |
| FrontierMath (tier 4) | 46.3% | 46.3% |
| Coding | ||
| Terminal-Bench 4.0 | 36.4% | 10.6% |
| LiveBench Agentic Coding | 57.2% | 64.1% |
| LiveBench Coding | 77.9% | 81.1% |
| Code Migration | 44.6% | 27.6% |
| CyberBench v1.1 | 75.5% | 69.4% |
| IOI | 47.3% | 43.9% |
| ProgramBench | 2.5% | 0.5% |
| Vibe Code Bench v1.1 | 90.4% | 82.9% |
| FrontierSWE | 43.8% | 25.8% |
| Agents | ||
| APEX-Agents | 56.1% | 58.6% |
| Terminal-Bench-Science | 10.0% | 4.3% |
| DTBench | 95.0% | 96.5% |
| LMCA | 51.1% | 53.9% |
| Mystery game puzzles | 30.0% | 25.0% |
| Professional work | ||
| Finance Agent (v2) | 54.1% | 58.9% |
| EMB | 59.0% | 62.7% |
| Legal Research Bench | 43.8% | 40.9% |
| Tax Agent Bench | 31.9% | 36.0% |
Tests at least two of these models have taken. Bold rows are the core tests.
