Verified model releases, product launches and material updates. Original short summaries, with the source one click away.
No new announcement verified for today yet. Recent updates are below.
OpenAI added an Ultrafast service tier for GPT-6.1 Sol in the Responses API. It speeds up output-token generation and supports global processing plus US and EU data residency. Access remains subject to rate limits and separate Ultrafast pricing.
Anthropic released Haiku 5.5 for inexpensive summaries, classification and other narrowly scoped tasks. It adds adjustable effort and is available across major cloud platforms. Pricing changes at prompts above 100,000 tokens; Anthropic still positions Sonnet and Opus for complex coding.
Microsoft announced a downloadable, roughly 3-bit version of MAI-Code-1.1-Flash retaining a 256K context window. Local model calls carry no inference charges, but Microsoft recommends over 120GB RAM. Experimental Copilot integration is planned by month-end; the cloud model already runs in Copilot.
OpenAI began a global ChatGPT Chat rollout combining GPT-6 with interactive visual responses. Paid tiers receive Sol; Free and Go receive Luna, beginning the following day. Enterprise access depends on admin settings. This announcement does not change Work or Codex models.
OpenAI introduced a dedicated Decisions API for turning text and images into typed answers. The public beta currently supports GPT-6 Luna only. It is aimed at decision-style tasks rather than a general replacement for the Responses API; the launch appears in the October 6 changelog.
Google announced Gemini 4 Argon for long-horizon engineering, professional work and defensive cybersecurity. Initial access is limited to trusted cyber defenders through Fairwind. Google says broader developer, enterprise and consumer access will follow after further safeguard testing; this is not a general-availability launch.
Anthropic released Sonnet 5.5 for well-scoped development and document work. It reports over 30% faster output and up to 30% lower task costs than Sonnet 5 at unchanged standard token prices. Those savings are vendor measurements, not guaranteed results for every workload.
Meta introduced technology that synchronizes expressive animated characters with Muse Realtime Voice. Generated video carries an invisible Video Seal watermark. The research examples demonstrate capability and do not all represent avatars offered in the app; Muse is intended for users aged 18 and older.
SpaceXAI released Grok 4.7 with a larger base model, training focused on longer tasks and improved self-checking. The announcement retains Grok 4.6 pricing: $2 per million input tokens and $6 output. Published benchmark results are the provider's reported measurements, not a universal ranking.