Model comparison
Muse Spark vs Qwen3-4B
Muse Spark is the stronger model overall, scoring 50.6 to 31.9 on the Noometry Index.
Last verified . 2 shared benchmarks.
Meta
Alibaba (Qwen)Summary
- They share 2 benchmarks with published results for both. Muse Spark scores higher in 3 categories and Qwen3-4B in 0 categories; 3 gaps are clear of the uncertainty.
- The widest gap is in knowledge, where Muse Spark leads 65.7 to 33.0.
- The biggest single-benchmark swing is GPQA Diamond: 89.8% for Muse Spark and 52.3% for Qwen3-4B.
- Qwen3-4B has downloadable open weights; the other is API-only.
Side by side
| Muse Spark | Qwen3-4B | |
|---|---|---|
| Provider | Meta | Alibaba (Qwen) |
| Noometry Index | 50.6 | 31.9 |
| Released | 2026-04-08 | 2025-04-29 |
| Weights | Proprietary | Open |
| Context window | — | — |
| Max output | — | — |
| Input $ / M tokens | — | — |
| Output $ / M tokens | — | — |
| Results tracked | 27 | 6 |
Sponsored placements are available on pages like this one. Advertise on Noometry
Category by category
Coding Not comparable
Muse Spark: 46.2 (#69), Qwen3-4B: —
| Benchmark | Muse Spark | Qwen3-4B |
|---|---|---|
| SciCode | 51.5% | — |
| LMArena Coding | 1481 | — |
Agentic & Tool Use Not comparable
Muse Spark: —, Qwen3-4B: 27.6 (#100)
| Benchmark | Muse Spark | Qwen3-4B |
|---|---|---|
| Berkeley Function Calling Leaderboard | — | 35.7% |
Reasoning Muse Spark leads
Muse Spark: 35.9 (#67), Qwen3-4B: 19.2 (#268)
| Benchmark | Muse Spark | Qwen3-4B |
|---|---|---|
| CritPt | 11.3% | — |
| Chess Puzzles | — | 4% |
| LMArena Hard Prompts | 1474 | — |
| Epoch Capabilities Index | 152.04 | — |
Math Muse Spark leads
Muse Spark: 47.8 (#66), Qwen3-4B: 29.7 (#240)
| Benchmark | Muse Spark | Qwen3-4B |
|---|---|---|
| OTIS Mock AIME 2024-2025 | 88.9% | 52.2% |
| MathArena Final-Answer Competitions | — | 38.5% |
| ProofBench | 17% | — |
| LMArena Math | 1455 | — |
| FrontierMath (Feb 2025 set) | 39% | — |
| FrontierMath Tier 4 (v1) | 14.6% | — |
Knowledge Muse Spark leads
Muse Spark: 65.7 (#13), Qwen3-4B: 33.0 (#208)
| Benchmark | Muse Spark | Qwen3-4B |
|---|---|---|
| GPQA Diamond | 89.8% | 52.3% |
| Humanity's Last Exam | 40.6% | — |
| Vectara Hallucination Rate | — | 5.7% |
| LMArena Expert | 1457 | — |
Multimodal Not comparable
Muse Spark: 43.4 (#24), Qwen3-4B: —
| Benchmark | Muse Spark | Qwen3-4B |
|---|---|---|
| LMArena Vision | 1306 | — |
| LMArena Document | 1444 | — |
Multilingual Not comparable
Muse Spark: 56.1 (#24), Qwen3-4B: —
| Benchmark | Muse Spark | Qwen3-4B |
|---|---|---|
| LMArena Non-English | 1464 | — |
| LMArena Chinese | 1509 | — |
| LMArena French | 1497 | — |
| LMArena German | 1497 | — |
| LMArena Korean | 1459 | — |
| LMArena Russian | 1466 | — |
| LMArena Spanish | 1472 | — |
Instruction Following Not comparable
Muse Spark: 75.9 (#51), Qwen3-4B: —
| Benchmark | Muse Spark | Qwen3-4B |
|---|---|---|
| LMArena Instruction Following | 1442 | — |
Long Context Not comparable
Muse Spark: 44.4 (#69), Qwen3-4B: —
| Benchmark | Muse Spark | Qwen3-4B |
|---|---|---|
| LMArena Longer Query | 1451 | — |
Writing & Preference Not comparable
Muse Spark: 66.0 (#39), Qwen3-4B: —
| Benchmark | Muse Spark | Qwen3-4B |
|---|---|---|
| LMArena Text | 1474 | — |
| LMArena Creative Writing | 1459 | — |
| LMArena Multi-Turn | 1477 | — |
Frequently asked questions
Is Muse Spark better than Qwen3-4B?
Muse Spark is the stronger model overall, scoring 50.6 to 31.9 on the Noometry Index.
How many benchmarks do Muse Spark and Qwen3-4B share?
2 benchmarks have published results for both models. Muse Spark has 27 scored results on Noometry and Qwen3-4B has 6.