Meta Rolls Out Muse Spark 1.3 With Deep Discounts for Data-Sharing Customers

  • Meta launched Muse Spark 1.3, a million-token coding model for long-context and agentic workflows.
  • Contributor pricing cuts input tokens to $0.10 and output tokens to $0.20 per million, with cached input at $0.002 per million.
  • Customers accepting data sharing receive lower prices but face a 60-request-per-minute limit, compared with 3,000 for standard users.

Meta has launched Muse Spark 1.3, a coding-focused AI model with a one-million-token context window and support for long-horizon, agentic workflows. Standard API pricing is $1.25 per million input tokens and $4.25 per million output tokens, compared with $0.10 and $0.20, respectively, for the Contributor tier, which requires customers to allow Meta to use prompts, responses and usage patterns in its training pipeline. Contributor cached-input pricing falls from $0.15 to $0.002 per million tokens, but throughput is limited to 60 requests per minute versus 3,000 for standard customers. The model is available through Muse Code, Meta's API and OpenRouter. Meta has not decided whether to release its weights, although it still plans to open-source Muse Spark 1.2's weights. The pricing strategy, first tested with Muse Spark 1.2 in August 2026, seeks coding data that Meta's consumer platforms generate less often and increases price competition with OpenAI and Anthropic.

The information on this website is generated using AI and we cannot guarantee its accuracy. Please use it as reference information only.