Meta Releases Muse Spark 1.3: 1M Multimodal Reasoning Model for Autonomous Agent Workflows
Meta Superintelligence Labs has published Muse Spark 1.3, an open-weights frontier agent model with 1,048,576 context tokens and comprehensive multimodal comprehension.
| Platform | Route Identifier | Context | Max Output | Rate / Tier | Status |
|---|---|---|---|---|---|
| OpenRouter | meta/muse-spark-1.3 | 1.05M tokens (1,048,576) | 944K tokens (943,718) | $1.25 / $4.25 per 1M tokens ($0.15 cached prompt) | active |
| Hugging Face Weights | meta-llama/Muse-Spark-1.3 | — | — | Free Open Weights (Self-hosted) | free |
Key Takeaways
- check_circleMeta Muse Spark 1.3 provides full 1M context capabilities with an open-weights license, permitting enterprise self-hosting and private cluster execution.
- check_circleSupports end-to-end multimodal reasoning across audio, video, image, and text simultaneously.
- check_circleCloud inference available on OpenRouter at $1.25/$4.25 per million tokens with $0.15 cache reads.
- check_circleAIMI logged the live route on OpenRouter on 2 September 2026 at 21:45:59 SAST.
Frontier Agent Reasoning in Open Weights
Meta continues its commitment to open science with Muse Spark 1.3. Sized as an advanced MoE architecture, Muse Spark 1.3 is specifically trained on long-horizon agent trajectories, terminal commands, and multi-file code editing.
Enterprises with strict data residency regulations can deploy the model weights on on-premise clusters or sovereign cloud infrastructure without sending proprietary code to third-party endpoints.
Frequently Asked Questions
Can I host Meta Muse Spark 1.3 on private hardware?
Yes, the weights are distributed under Meta's permissive commercial license on Hugging Face for private server and cloud deployments.
Related Frontier Models & Releases
Explore verified specifications, benchmark results, and route pricing across alternative models in this class.
Meta Muse Glimmer 30B: An Open-Weight Agentic Model Built to Run Locally
Meta Superintelligence Labs has released Muse Glimmer 30B, an Apache 2.0 open-weight model distilled from Muse Spark and tuned for local agent workflows on consumer GPUs. It covers tool use, long-context reasoning and multimodal input in a package that fits under 20 GB with 4-bit quantisation.
Alibaba Qwen Releases Qwen3.8 Max: 2.4-Trillion Parameter Flagship with 1M Multimodal Context
Alibaba's Qwen team has launched Qwen3.8 Max, a 2.4T parameter Mixture-of-Experts model offering 1M token context, native video perception, and deep agent tool orchestration.
Moonshot AI Releases Kimi K3: 2.8T Parameter Open-Weight Multimodal Frontier Model
Moonshot AI has released Kimi K3, a 2.8-trillion parameter MoE powerhouse featuring 1M context, native video understanding, and leading mathematical reasoning.