Meta
Muse Spark Open
Meta's open-weight reasoning model for coding and agents, with a million-token context window for working across large inputs.
Run this model
meta/muse-spark-open
Use an endpoint supported by this model.
Get a key for this setup →const response = await fetch('https://api.minirouter.sh/v1/chat/completions', { method: 'POST', headers: { Authorization: `Bearer ${process.env.MINIROUTER_KEY}`, 'Content-Type': 'application/json', }, body: JSON.stringify({ model: 'meta/muse-spark-open', messages: [{ role: 'user', content: 'Why is the sky blue?' }], max_tokens: 1024, }),}) const data = await response.json()console.log(data.choices[0].message.content)Start with text input and an explicit max_tokens limit. Leave room for reasoning and the answer within the context window.
Model overview
- Input / 1M
- $1.3125
- All-in MiniRouter rate
- Output / 1M
- $4.4625
- All-in MiniRouter rate
- Context
- 1M
- 1,048,576 tokens
- Max output
- 1M
- tokens
- Architecture
- 816B / 42B
- Total / active parameters per token
- Reasoning
- Low through xhigh
- Four effort levels
- Announced
- 15 Oct 2026
- publisher release
Sparse mixture of experts
Large capacity. A smaller active slice.
Muse Spark Open activates about 42 billion of its 816 billion parameters per token. Meta positions it for code changes, multi-step reasoning and long-running agents.
Meta architecture
- Total parameters
- 816B
- Active per token
- 42B
View chart data
| Parameters | Billions |
|---|---|
| Total | 816 |
| Active per token | 42 |
API and open weights
Choose the access that fits your work.
Use the hosted API for text workflows. The downloadable weights offer a separate path for running and adapting the model yourself.
- [ 01 ]
Start with text
Send a prompt, choose a reasoning level and set an output budget.
- [ 02 ]
Compare completed tasks
Try repository analysis, debugging or a bounded code change. Check correctness, latency and total cost before switching a production workload.
- [ 03 ]
Run your own model
Download the weights to adapt and host the model under the Muse Model License.
Route
One route, standard clients.
Authenticate with a MiniRouter bearer key. Caller-provided upstream credentials are not used on this route.
Base URL
https://api.minirouter.sh/v1Model ID
meta/muse-spark-openEndpoints
/v1/chat/completions/v1/messages/v1/responsesParametersmax_tokens, temperature, stop, tools, tool_choice, reasoning, include_reasoning
Structured outputNot enforced; a requested format is dropped and the answer is plain text.
ModalitiesInput: text · Output: text
Configured providers. Priority follows the live catalog order.
| Provider | Priority | Input / 1M | Output / 1M | Parameters | Access |
|---|---|---|---|---|---|
vercel | 1 | $1.3125 | $4.4625 | max_tokens, temperature, stop, tools, tool_choice, reasoning, include_reasoning | MiniRouter balance |