Muse Spark 1.2
Meta · released Aug 5, 2026
- Type
- Closed
- Input
- $1.25
- Output
- $4.25
- Cached
- $0.15
List price · per 1M tokens · Meta at 1M context · source ↗
Our take
Written Sep 17, 2026Muse Spark 1.2 is a hosted-only model from Meta, so using it means choosing a host rather than running it yourself. Its measured strengths sit in maths and reasoning, and it takes text, images, files, audio and video in the same request.
Reach for it on maths and reasoning work, where it scores 91.2% and 90% on the LiveBench categories measured 2026-06-25, or when a long document has to go into one request without being split first. Skip it if you need to run the model on your own machine, or if you need agent-driven coding fixes landed without review.
The case for it
- Maths and reasoning are its highest measured categories, at 91.2% and 90% on LiveBench (evaluated 2026-06-25), so it suits analytical work you can check.
- Text, images, files, audio and video all go into the same request, so a screenshot or a recording does not have to be described in words first.
- The request capacity takes a long report or a stack of documents beside the question, though reliable recall across all of it is unverified in our data.
The case against it
- Agentic coding is its weakest measured area at 57.58% on LiveBench (2026-06-25), against 77.54% on its general coding category, so set-piece code generation is the safer use.
- We list no download for it, so using it means choosing a host; no licence is supplied in our data, so permissions are unverified.
- The parameter count is not disclosed, so its size and memory footprint cannot be assessed.
How good is it?
A closed text model for everyday questions, coding and drafting, though it can be slow to change course when you give new instructions.
- getting answers to everyday questionsArena Text (overall) · 5th of 168
- drafts, rewrites and editingArena Creative Writing · 28th of 168
- writing and completing codeArena Coding · 8th of 168
- getting back on track after a step failsArena Agent · Recovery · 12th of 55
- changing course when you give new instructionsArena Agent · Steerability · 43rd of 55
EverydayGeneral questions and everyday reasoning
Arena Text (overall)5th of 168 · 1496
CodingWriting and fixing code on its own
Arena Coding8th of 168 · 1535
AgenticPlanning, calling tools, staying on task
Arena Agent37th of 55 · −0.03
WritingDrafting and rewriting prose
Arena Creative Writing28th of 168 · 1448
Placings on Arena's agent boards, from live sessions people ran themselves. A model can lead on one of these and sit mid-field on the others.
Boards this model appears on that none of the ratings above are built on.
Every published score for this model19 scoresEvery figure we hold, from 19 boards, with who ran it and a link to the source — including the boards no rating above is built on.
Where to rent it
Prices checked 1 hour ago — each listing carries its own date.
Meta, through OpenRouter
The lab is also the cheapest we hold. The strip above and this offer are the same one, so nothing on this page undercuts Meta on 1M of context.
- per 1M tokens
- $1.25 in / $4.25 out
- Context served
- 1M
- Throughput
- ~181 tok/s
| Provider | In / out per 1M tokens | Context | Throughput | Trains on prompts | Logs prompts | Zero retention |
|---|---|---|---|---|---|---|
| OpenRouterOpenRouter's own listing | $1.25 / $4.25checked 1 hour ago | 1M | not measured | Unknown | Unknown | Unknown |
| MetaThrough OpenRouter | $1.25 / $4.25checked 1 hour ago | 1M944K max reply | 181 tok/s | No | Yes30 days | Unknown |
OpenRouterAcross the 2 listings we hold: 1 says it does not train on prompts, 0 say they do and 1 does not say. 0 appear in the zero-retention registry we check; the rest are unknown to us.
What each host's API supports
From the parameter list each endpoint publishes. Streaming is omitted: nothing we hold reports it, for any model.
| Provider | Tool calling | JSON output | Strict schema |
|---|---|---|---|
| OpenRouterOpenRouter's own listing | ✓ | ✓ | ✓ |
| MetaThrough OpenRouter | ✓ | ✓ | ✓ |
Tool calling: 2 of 2 listings say yes. JSON output: 2 of 2 listings say yes. Strict schema: 2 of 2 listings say yes.
Models people weigh against Muse Spark 1.2
When we formed this view
Recent changes
Each date is the day we first saw the change, or the day the maker announced it.
What we do not know about this model yet
- Nothing we hold says whether an endpoint streams, so we do not show it either way.
- 1 of 2 listings does not say whether it trains on prompts.
- We hold no batch or off-peak rate for any of its listings.
Licence and identifiers
What the licence allowsWe hold no licence record for this model. Inside are the identifiers you need to pull it — its Hugging Face repo where we have one, our slug and a machine-readable card.
Licence
We hold no licence record for this model, and no record of published weights either — so we can neither summarise its terms nor point you at the weights.
Identifiers
- Takes in, gives back
- Text, images, audio, video and documents in, text out
- Catalogue slug
- meta-muse-spark-1-2