Muse Spark 1.3
Muse Spark 1.3
Meta's closed model has caught up with the frontier, and its cheapest tier is paid for with your data.
Updated
The read
Muse Spark 1.3 is the third update since Meta launched its closed line, and the first that genuinely belongs near the top: ninth on the arena leaderboard and within a few points of Claude Opus 5 on the Artificial Analysis index. It reads text, images, video and PDFs over a million-token window. The standard price is reasonable, and a much cheaper contributor tier exists if you let Meta train on what you send. The open-weights release Meta keeps promising has not arrived.
Where it shines
- Ninth on the arena leaderboard as of October
- Million-token window across text, images, video and PDFs
- A very cheap tier for work you are happy to share
Worth knowing first
- The cheap tier lets Meta train on your data
- Closed weights, despite Meta's open-source history
Index score
85.7
Provisional. This score is read from the product's public capability, not from the one-prompt rebuild the fully reviewed entries went through, so treat it as a placing rather than a verdict.
- Rank
- 16 of 73
- Builds
- LLMs
- Category
- Frontier
- Output
- Text, images, video and PDFs in, text out
- Pricing
- Paid only
- Best for
- Multimodal work for teams comfortable inside Meta's platforms
Scores are our own editorial judgement, weighted craft 55, speed 10, control 15 and value 20.
Also worth a look
Full index- 01Claude Opus 5.5Claude Opus 5.5ProvisionalFable-level answers for less than the Opus it replaces. The new default.Frontier modelPaid only92.8
- 03Gemini 4 ArgonGemini 4 ArgonProvisionalTop of the human-preference leaderboard, and almost nobody can use it yet.Frontier modelPaid only90.6
- 04GPT-6.1 SolGPT-6.1 SolProvisionalNear-Astra answers at a fifth of the price. The OpenAI model most teams should deploy now.Frontier modelPaid only90.3
