Meta: Muse Spark
28 benchmarks
Leaderboard position
#69
Overall rank (of 1126 models)
2500
Elo rating
28
Benchmarks
Top 6.1%
Percentile
In Meta (of 12 models)
Above: OpenAI: GPT Image 1.5 #68
Below: Qwen: qwen3.8-flash-next #70
Benchmark highlights
Best benchmark
Text Arena (Snapshot from 20 june 2026)
1487
Worst benchmark
DeepSWE 1.1
10
Nearest competitor from another company
The model from another company with the most similar benchmark scores.
appeared as nearest in 2 of 28 benchmarks
2/28
7.1%
2/28
7.1%
| Model | Votes | % of benchmarks |
|---|---|---|
| Ornith: ornith-1.5-35b-a3b | 2 | 7.1% |
| Moonshot AI: kimi-k3 | 2 | 7.1% |
| NVIDIA Corporation: nvidia-nemotron-3.5-super-vl-120b-a12b | 2 | 7.1% |
| Qwen: qwen3.6-max | 1 | 3.6% |
| MiniMax: minimax-m2.7 | 1 | 3.6% |
| Anthropic: Claude Opus 4.7 | 1 | 3.6% |
| DeepSeek: DeepSeek V4 Flash | 1 | 3.6% |
| nanbeige: nanbeige4.2-3b | 1 | 3.6% |
| Qwen: qwen3.8-2.4t-a95b | 1 | 3.6% |
| Z.ai: GLM 5.2 | 1 | 3.6% |
Benchmarks
Text Arena (Snapshot from 20 june 2026)
1487
TauBench V2 (Telecom)
92
ChartQA
88.9
GPQA Diamond
88
MCP-Atlas
82.2
MMMU-Pro
81
IFBench
76
AA-LCR
70
Terminal-Bench v2.1
67.3
Artificial Analysis Agentic Index (Maximum Reasoning)
62
SWE-bench Pro
55
OSWorld-Verified
53.3
Artificial Analysis Intelligence Index v4.0 (Maximum Reasoning)
52.2
SciCode
52
Humanity's Last Exam (with tools)
50.4
Toolathlon Verified
49.4
AA-Omniscience Accuracy
45
Terminal-Bench Hard
45
Artificial Analysis Intelligence Index v4.1 (Maximum Reasoning)
43
HealthBench Hard
42.8
Humanity's Last Exam (no tools)
40
BabyVision
39.9
GDPval-AA v2
33
AA-Omniscience Non-Hallucination Rate
27
TauBench V3 (Banking)
20
JobBench
17
CritPt
11
DeepSWE 1.1
10
Predicted scores
Estimated from 25 similar models and the global Elo rating. These are projections, not measured results.