Meta Ships Muse Spark 1.1, Its First Paid Agentic Model

Source: MarkTechPost — 2026-07-09

Summary

Meta Superintelligence Labs released Muse Spark 1.1, its second model and first paid offering through a new Meta Model API. Positioned as a frontier-tier coding/agentic model with a 1M-token context window, parallel tool calling, and computer-use support, it's Meta's clearest move yet toward closed, commercially-licensed models sitting alongside its open Llama line, directly targeting GPT-5.6, Claude, and Gemini in agentic coding workloads.

Key Takeaways

• Muse Spark 1.1 ships via a new paid Meta Model API — a first for Meta, whose flagship models have so far been open-weight (Llama).
• Supports a 1M-token context window, parallel tool calling, and computer-use, aimed squarely at agentic coding and long-horizon tasks.
• Signals Meta running two parallel model strategies: open Llama for community/ecosystem reach, closed Muse Spark for frontier commercial competition.
• Positions Meta as a fourth major closed-model competitor alongside OpenAI, Anthropic, and Google in the agentic-coding race.

Discussion

(No questions yet — ask follow-ups via a Claude Code chat session on this repo; answers get appended here.)