Meta publicly launched Muse Spark 1.1 on July 9, 2026, a multimodal AI model designed for agentic coding, with API access available through Meta's developer platform. This release positions Meta as a major competitor to OpenAI and Anthropic in the AI coding assistant space, offering aggressive pricing at $1.25 per million input tokens and potentially disrupting the market by commoditizing coding models. The model is evaluated on Terminal-Bench 2.1, but community analysis highlights that resource limits (6 CPU cores, 8GB RAM) were overridden, which may disqualify results. Pricing is $1.25/$4.5 per million tokens for input/output, with $0.15 for cached input.
Background
Muse Spark is Meta's most powerful AI model, first introduced in April 2026 as part of a broader scaling effort. Agentic AI systems can pursue goals, use tools, and take actions autonomously, going beyond traditional text generation. Version 1.1 specifically targets coding tasks with agentic capabilities.
References
Discussion
Community members are skeptical about benchmark validity; GodelNumbering points out that overriding resource caps in Terminal-Bench 2.1 disqualifies results. Simon Willison shared a practical integration tool for LLM, while others debate the strategic impact and pricing, with some noting that Meta's low prices could commoditize coding models.