Meta Superintelligence Labs, the unit built after Meta’s ~$14.3 billion acquihire of Scale AI led by Alexandr Wang, shipped its first commercial product on August 5, 2026: Muse Code, a terminal-based coding agent powered by the Muse Spark 1.2 model. The launch marks Meta’s split of research from product, with a revenue mandate on the product side. Muse Code installs on macOS and Linux with a single command and uses persistent asynchronous background agents for long-horizon tasks. On vendor-reported benchmarks, Muse Spark 1.2 scores 82.9% on Terminal-Bench 2.1 and 59.3% on DeepSWE 1.1, improvements of 6.7 and 6.3 points over Spark 1.1, trailing only Claude Opus 5 at 86.7% on Meta’s chart. Standard pay-as-you-go access costs $1.25 per million input tokens and $4.25 per million output. The key strategy is the contributor tier at approximately $0.10/$0.20 per million tokens, roughly 25% of what OpenAI and Anthropic charge at list. Developers on this tier opt in to having their prompts and completions used to improve Meta’s models, creating a feedback loop where subsidized access generates high-quality coding data for the next training run, turning the user base into a distributed data-generation engine.