Meta/muse-spark-1-3/Released Sep 2026

Muse Spark 1.3

Agentic coding with 1M context, fewer tool calls

Text

About Muse Spark 1.3

Muse Spark 1.3 is Meta's multimodal reasoning model for long-running agentic, multi-agent, and coding workflows. It can plan across tools, delegate work to parallel subagents, work with large codebases, and sustain longer-horizon tasks while juggling multiple workflows in a single extended thread.

Compared with earlier Muse Spark releases, it completes engineering tasks with fewer tool calls and fewer output tokens, takes fewer turns where they are not needed, and produces a cleaner coding style. Meta also reports stronger resistance to adversarial inputs and prompt injections. The model supports structured output, parallel tool calling, and visual understanding.

MetricValue
Context Length1,048,576 tokens
Max Output131,072 tokens

Ready to build with Muse Spark 1.3?

Try Muse Spark 1.3 in the Workbench to prompt it, compare outputs, and iterate on prompts without writing any code. When you're ready to ship, call the same model from our API and build your own apps on top of it.

Sign in to grab your API key and start making requests.
bash
curl -sSf -X POST https://hub.oxen.ai/api/ai/chat/completions \
    -H "Content-Type: application/json" \
    -H "Authorization: Bearer $OXEN_API_KEY" \
    -d '{
  "model": "muse-spark-1-3",
  "messages": [
    {
      "role": "user",
      "content": "Try sending a message."
    }
  ]
}'