Muse Spark Open is an 816B total / 42B active open-weight multimodal reasoning model from Meta for coding and long-horizon agentic work, with 1M-token context.
Muse Spark Open by meta is offered by 12 providers on this page. Its reference price is $1.56 per 1M input tokens and $5.31 per 1M output tokens. The lowest paid channel is Kilo Gateway at $1.25 / $4.25 per 1M, about 1.3× below the reference price. That channel is a third-party gateway, so confirm its availability and rate limits before depending on it.
Artificial Analysis rates it 38.9 on the Intelligence Index. Against its lowest blended price of $2.00 per 1M, that is roughly 19 index points per dollar, which is the value ratio the leaderboards rank on. Median output speed is 102 tokens per second, with 0.67s to the first token. Running one task of the Artificial Analysis suite costs about $0.8435, which reflects how many tokens its reasoning consumes rather than the unit price alone.
The context window is 1,048,576 tokens, with an output limit of 131,072 tokens. It supports reasoning, tool use, and structured output. Accepted input modalities are Text, Image, Video, PDF, and Audio. The weights are open, so it can also be self-hosted or served through a gateway of your choice. Its training knowledge cuts off at 2026-01-04.
Quality is independently evaluated by Artificial Analysis. Speed/latency are model-level medians.
| Provider | Tier | Input | Output | Cache read | Cache write | Context | Output limit | Status |
|---|---|---|---|---|---|---|---|---|
| Kilo Gateway | Gateway | $1.25 | $4.25 | $0.15 | — | 1,048,576 | 943,718 | |
| OpenRouter | Gateway | $1.25 | $4.25 | $0.15 | — | 1,048,576 | 943,718 | |
| Ollama Cloud | Cloud | $1.25 | $4.25 | $0.15 | — | 1,048,576 | 131,072 | |
| Nebius Token Factory | Cloud | $1.25 | $4.25 | $1.25 | — | 1,048,576 | 1,048,576 | |
| Vercel AI Gateway | Cloud | $1.25 | $4.25 | $0.15 | — | 1,048,576 | 1,048,576 | |
| Fireworks AI | Cloud | $1.25 | $4.25 | $0.15 | — | 1,048,576 | 131,072 | |
| Merge Gateway | Gateway | $1.25 | $4.25 | $0.15 | — | 1,048,576 | 262,144 | |
| Eden AI | Gateway | $1.25 | $4.25 | $1.25 | — | 1,048,576 | 131,072 | |
| Eden AI | Gateway | $1.25 | $4.25 | $0.15 | — | 1,048,576 | 131,072 | |
| Requesty | Gateway | $1.25 | $4.25 | $0.15 | — | 1,048,576 | 131,072 | |
| Prime Inference | Gateway | $1.25 | $4.25 | — | — | 1,048,576 | 943,718 | |
| Venice AI | Gateway | $1.56 | $5.31 | $0.19 | — | 1,048,576 | 131,072 |
Sorted by blended price (input×0.75 + output×0.25) asc. The official channel always shows regardless of rank. Whether a gateway's low price is actually usable can't be verified.
Interleaved thinking (reasoning between tool calls) is declared by 1 of 12 providers.
1 / 12 providers expose no reasoning control (reasoning_options: []).
Price history accumulates from each data sync; currently only 1 sample(s) (2026-10-16). Each future sync adds a point, and once accumulated a line is drawn here.
| Name | Conditions | Score | Metric | Source |
|---|---|---|---|---|
| Artificial Analysis Intelligence Index | variant: xhigh effort | 39 | index score | Source ↗ |
| JobBench | variant: xhigh effort | 61.8 | — | Source ↗ |
| MCP Atlas | variant: xhigh effort | 89.2 | — | Source ↗ |
| GDPval-AA | variant: xhigh effort · v2 | 1401 | Elo | Source ↗ |
| OSWorld-Verified | variant: xhigh effort | 80.4 | — | Source ↗ |
| Toolathlon-Verified | variant: xhigh effort | 72.8 | — | Source ↗ |
| HLE-Diamond | variant: xhigh effort, no tools | 26.1 | — | Source ↗ |
| AA-Omniscience | variant: xhigh effort | 25 | — | Source ↗ |
| SciCode | variant: xhigh effort | 58.5 | — | Source ↗ |
| Humanity's Last Exam | variant: xhigh effort, with tools | 61 | — | Source ↗ |
| GPQA Diamond | variant: xhigh effort | 89 | — | Source ↗ |
| DeepSWE | harness: preferred harness · variant: xhigh effort · v1.1 | 60 | — | Source ↗ |
| SWE-Atlas Codebase QnA | harness: preferred harness · variant: xhigh effort | 58 | — | Source ↗ |
| Terminal-Bench | harness: Terminus 2 · variant: xhigh effort · v2.1 | 76.4 | — | Source ↗ |
| ZeroBench | variant: xhigh effort, with tools | 50 | — | Source ↗ |
| BabyVision | variant: xhigh effort, with tools | 80.2 | — | Source ↗ |
| SimpleVQA | variant: xhigh effort, with tools | 73.6 | — | Source ↗ |
| WorldVQA | variant: xhigh effort, with tools | 49.4 | — | Source ↗ |
| IntPhys2 | variant: xhigh effort | 68.3 | — | Source ↗ |
| Video-MME | variant: xhigh effort · v2 | 61.5 | — | Source ↗ |
The same benchmark scores very differently across harness / dataset, so the qualifying conditions must be shown together.