Recent launches worth knowing about, each linked to the official announcement. We only list items we can trace to a primary source.
Meta says it is releasing the weights for Muse Spark Open, a natively multimodal open-weight model with a 1M-token context window, built for coding and agentic work. Meta lists availability through several API platforms and model hubs including Hugging Face.
Source: Meta (opens in a new tab)
NVIDIA announced Nemotron 3.5 Super VL, an open 120B-parameter model for always-on agents, alongside Nemotron 3.5 Bolt edge models for local agentic workloads. Super VL and Bolt 74B-A9B are listed as available now; Bolt E4B is listed as coming Oct. 23.
Source: NVIDIA (opens in a new tab)
Google says Gemini 4 Argon is globally available to developers and enterprises via the API in AI Studio, and to Google AI Ultra and Pro subscribers in the Gemini app and AI Mode in Search. Google describes a 1M-token limit and a focus on complex coding, scientific and agentic work.
Source: Google (opens in a new tab)
Meta launched Muse Wonder on its Muse Code agent and Meta Model API, and Muse Wonder Full exclusively through Muse Code subscription plans. Meta positions the models for coding, long-running agents and computer use.
Source: Meta (opens in a new tab)
Anthropic describes Claude Haiku 5.5 as its cheapest, fastest and most capable small model, aimed at high-volume, latency-sensitive work and as a subagent for coding. It is available on the Claude Platform and major clouds, and in Claude Code.
Source: Anthropic (opens in a new tab)
Shopify announced Canvas, a visual design surface where merchants build and customize online stores together with Sidekick, Shopify's AI agent. Shopify says it is rolling out to merchants over the coming days and is not yet replacing the existing editor.
Source: Shopify (opens in a new tab)
Summaries paraphrase the linked source and are not endorsements or rankings. Spotted something we missed? Tell us. Also available as an Atom feed.