Muse Spark 1.3: benchmarks, price and rank
Muse Spark 1.3 from Meta is 15th of 123 rated AI models on Spezia's leaderboard, strongest on MRCR v2 (8 needles) and LMArena text. It costs $1.25 in, $4.25 out per million tokens.
- 15th of 123 rated models
- Rating 1,700
- First on 1 of 10 benchmarks
- Closed
- Released Sep 2, 2026
- 1 million tokens of context
Where it is strongest
- 1st of 24 on MRCR v2 (8 needles)
- 10th of 107 on LMArena text
Where it falls behind
- 15th of 21 on FrontierSWE
- 14th of 20 on CursorBench
Price: $1.25 in, $4.25 out per million tokens.
Its headline results
The benchmarks that count most in the rating, with its place among the models measured the same way, and where each number came from.
- APEX-Agents58.6%
Long tasks written by investment banking analysts, management consultants and corporate lawyers from their own work, done across files and apps in a realistic work setup.
20th of 56. Independent result from Epoch AI.
- FrontierSWE25.8%
Open-ended technical projects that take hours to tens of hours, from making systems faster to building large programs and applied machine learning research. Each run gets a 20-hour budget.
15th of 21. Independent result from Epoch AI.
- CursorBench41.6%
Ambiguous coding tasks that span many files, taken from real sessions in the Cursor code editor.
14th of 20. Independent result from Epoch AI.
- DeepSWE 1.175.4%
Software engineering tasks run in the same simple agent setup for every model.
4th of 27. The maker's own number, from Meta, via LLM Stats.
- SimpleBench81.8%
Multiple-choice puzzles about everyday situations, space, time and people, written to trip up models that pattern-match instead of think. Ordinary adults do well on them.
10th of 68. Independent result from Epoch AI.
- FrontierMath74.4%
Hundreds of new, unpublished math problems written by mathematicians, from hard undergraduate level up to research, with answers a computer can check.
26th of 76. Independent result from Epoch AI.
- MRCR v2 (8 needles)98.5%
A very long conversation full of similar requests, where the model has to find and repeat particular earlier answers, eight of them at once.
1st of 24. The maker's own number, from Meta, via LLM Stats.
- LMArena text1,491
People put a question to two anonymous models and vote for the better answer.
10th of 107. Independent result from LMArena.
- LMArena WebDev1,655
People ask two anonymous models to build the same web page or app and vote for the better one.
17th of 98. Independent result from LMArena.
- DeepSearchQA89.4%
Questions that need many searches and several steps of reasoning across sources to answer.
4th of 8. The maker's own number, from Meta, via LLM Stats.
2 more results in its full profile.
Every result with its source, what it costs per task, what jumped with its release, side by side with any other model, and answers to your questions about when to use it.
Models ranked near it
How the rating works, and every benchmark it uses: the model leaderboard.