AIBRIEF
Meta's AI coding agent Muse Code launched August 5 with output priced at 20 cents per million tokens of AI generated text for users sharing feedback — matching China's DeepSeek and undercutting OpenAI's discounted older models.
Meta launched Muse Code, an AI coding agent, on August 5 with a discounted tier that prices output at 20 cents per million tokens for users willing to share feedback with the company (Yahoo Finance). The default tier matches Meta's general Muse Spark model at $1.25 per million input tokens and $4.25 per million output tokens (CNBC, Engadget).
The 20-cent floor lines up with China's DeepSeek and undercuts OpenAI's discounted older models. On Terminal-Bench 2.1, Muse Code ranked second, behind Anthropic's Claude Code Opus 5 and ahead of OpenAI's Codex in its first release (eWeek). Meta AI chief Alexandr Wang said the pricing can be an incredibly good option for a lot of workflows, especially from a cost perspective (Yahoo Finance).
The launch lands as investors press for proof that Meta's AI spending produces revenue: shares fell about 10% the week before after Zuckerberg's earnings call. The same week, Meta disclosed that a separate model, Muse Spark 1.1, had a contained security incident during third-party testing by evaluator Irregular; a misconfiguration gave it unintended internet access (Yahoo Finance). Muse Code and Muse Spark 1.1 are different products.
What remains unknown: whether the 20-cent feedback tier becomes the category floor, or stays a loss leader aimed at developer mindshare.