Safe Superintelligence reportedly preparing to release a model in August, and Meta launches Muse Code coding agent

Safe Superintelligence and Meta Move on August Releases

Safe Superintelligence is preparing to release a model this month, according to reports cited in the segment. The timing remains unconfirmed in the transcript, but the news signals continued pressure in the coding-agent space as multiple players move toward product launches simultaneously.

Meta launched Muse Code in beta today—a terminal coding agent that handles complete software engineering tasks. The product runs on Llama Spark 1.2, Meta's coding-focused model update. Mark Zuckerberg announced the release on X, positioning it as a direct competitor in the agentic coding market.

The competitive dynamic is tightening. Microsoft has meanwhile standardized on OpenAI's GPT 5.6 Sole as its internal default model following an exec memo designed to end token maxing. The move signals that model preference—not just availability—is now table stakes for enterprise adoption.

Distillation remains a tension point. Anthropic has publicly accused multiple labs of distilling Claude, including DeepSeek, Moonshot, MiniMax, ZAI, and Alibaba. The company's stated position is willingness to sacrifice short-term gains for longer-term IP protection, though it's unclear whether that commitment will hold if distillation attacks persist or intensify.