LLM
Meta Muse Spark 1.1 Review
Meta Superintelligence Labs · LLM
Meta Muse Spark 1.1 launched July 9, 2026 from Meta's Superintelligence Labs. It's Meta's most capable model to date, purpose-built for agentic workflows:
Meta Muse Spark 1.1 launched July 9, 2026 from Meta's Superintelligence Labs. It's Meta's most capable model to date, purpose-built for agentic workflows: it acts as a main orchestrator that delegates subtasks to parallel sub-agents to reduce latency. Key specs: 1-million-token context window with active memory management, strong computer-use capabilities (it can directly control UIs and run automation scripts), and a best-in-class price of $1.25 per million input tokens / $4.25 per million output tokens via the Meta Model API. New accounts get $20 in free credits. Available in both standard and Thinking modes on meta.ai.
Pros
- Cheapest frontier-grade agentic model at $1.25/M input tokens
- 1-million token context with active memory
- Multi-agent orchestration built into the architecture
- Computer use: can control UIs directly
- $20 free credits for new API accounts
Cons
- Meta Model API still in US-only public preview
- No published comparative benchmarks vs GPT-5.6 or Claude
- Less enterprise tooling than OpenAI or Anthropic
- Thinking mode not available via API yet
Use cases
- Long-horizon coding projects (full feature implementation)
- Agentic browser automation and data extraction
- Cost-sensitive high-volume API workloads
- Research tasks requiring million-token context
- Multi-agent pipeline orchestration
Pricing
freemium — from $1.25/M tokens. A free tier is available.