Models & Research
Meta launches Muse Wonder and Muse Wonder Full for agentic tasks
Meta introduced Muse Wonder and Muse Wonder Full, two new frontier models aimed at personal and professional agentic work. According to Meta, the models are designed for tasks such as navigating insurance claims, understanding leases, and helping build apps, and they add features Meta calls Active Memory and Active Focus to better sustain multi-step work over time. Meta says Muse Wonder is available now to developers and builders through the Meta Model API and Muse Code, while Muse Wonder Full is available now to Muse Code subscribers. Meta also says the models will soon power more experiences in Muse, starting with subscribers.
Anthropic details multi-day Claude workflows for scientific computing
Anthropic published a research post on long-running Claude for scientific computing, describing how to structure agentic coding work that can run across many sessions instead of a single short chat. The piece focuses on practical workflow design: using test oracles to check progress, keeping persistent memory so the agent can resume effectively, and orchestrating sub-tasks when debugging or exploration becomes too complex for one straight-line pass. Anthropic says the patterns are meant for scientific computing, but the guidance is presented as broadly useful for other long-running agentic coding projects as well. The post also points to an internal C compiler project as an example of Claude operating across roughly 2,000 sessions. The development is best read as a workflow and orchestration playbook, not a new model release.
Anthropic launches Claude Haiku 5.5 as a cheaper, faster small model
Anthropic announced Claude Haiku 5.5 on October 7, 2026, positioning it as the company’s cheapest, fastest, and most capable small model so far. The model is aimed at high-volume, cost-sensitive work such as summaries, classification, compaction, and subagent tasks in coding workflows. Anthropic also says Haiku 5.5 is available now across its own platform and major cloud providers. In its launch post, the company reports benchmark gains over Haiku 4.5 and says the new model includes an adjustable effort setting. Anthropic additionally says it cut Sonnet 5.5 cache-read pricing and introduced new monthly API credits for Max and Team subscribers. These are Anthropic’s product and performance claims, not independently verified measurements.