Muse Spark 1.3 is a multimodal reasoning model from Meta for long-running agentic, multi-agent, and coding workflows. It improves long-horizon agent collaboration, instruction following, and coding efficiency relative to Muse Spark 1.2.
Specification
most-agreed values
Context
1M
Max output
131K
Released
2026-09-02
Knowledge cutoff
—
Retires
—
Open weights
no
Input
text, image, video, pdf, audio
Output
text
Price
US dollars per million tokens · most-agreed
Input
$1.25
Output
$4.25
Cache read
$0.15
Quality
1 benchmark · 1 source
Composite
97th
percentile of 334 scored models
Rank
12 / 334
±0.53 sd
Evidence
1 × 1
too little evidence to place with confidence
Effort range
—
one configuration only
general95th1/1 bench
Every score
one row per source and configuration — nothing averaged away
| Benchmark | Score | Configuration | Source | Run |
|---|---|---|---|---|
| LiveBench | 82.4 | xhigh · 23 runs | LiveBench | 2026-06-25 |
The leading figure for a benchmark is the median across its configurations, so one heroic high-effort run cannot set the number.
Available from 9 hosts
| Host | In $/M | Out $/M | Cache rd | Cache wr | Context | Output | Retires |
|---|---|---|---|---|---|---|---|
| empiriolabs | $1.25 | $4.25 | $1.00 | — | 1M | 131K | — |
| kilo | $1.25 | $4.25 | $0.15 | — | 1M | 944K | — |
| llmgateway | $1.25 | $4.25 | $0.15 | — | 1M | 131K | — |
| llmgateway-providers | $1.25 | $4.25 | $0.15 | — | 1M | 1M | — |
| meta | $1.25 | $4.25 | $0.15 | — | 1M | 131K | — |
| nano-gpt | $1.25 | $4.25 | $0.15 | — | 1M | 944K | — |
| opencode | $1.25 | $4.25 | $0.15 | — | 1M | 131K | — |
| openrouter | $1.25 | $4.25 | $0.15 | — | 1M | 944K | — |
| vercel | $1.25 | $4.25 | $0.15 | — | 1M | 1M | — |
More from Meta
most-hosted first