Kimi K3.1 Pro
Moonshot AI
First model in the K3.1 series, plus a Highspeed variant. Moonshot describes a 3T-class open-weight model with a 1M-token context, tuned for coding and agentic work.
Official source ↗New AI tool releases (October 2026)
Notable AI launches from October 2026, each with its official source. If we couldn't verify the date, it's not here. Updated .
Claude Duet (research preview)
Anthropic
A new app where Claude works as a proactive agent: surfaces what needs your attention, keeps working in the background and asks before acting. Waitlisted preview for US Pro and Max users.
Official source ↗Voxtral 3
Mistral
Speech-to-text and text-to-speech model family with a voice-agent SDK, via Mistral's EU-hosted API or self-deployment.
Official source ↗Muse Spark Open
Meta
Open-weight multimodal model (816B total / 42B active, 1M context) aimed at coding and agentic work.
Official source ↗Grok Voice 3.0
xAI
New speech-to-speech model behind xAI's Voice Agent API; the grok-voice-latest alias now points to 3.0.
Official source ↗Gemini 4 Argon (broader access)
Google
Long-horizon reasoning model for software engineering and knowledge work, now in the paid Gemini API, Android Studio, and for eligible Gemini subscribers.
Official source ↗Claude Haiku 5.5
Anthropic
Anthropic's cheapest, fastest small model, for high-volume tasks and coding subagents, with an adjustable effort setting.
Official source ↗Nano Banana 2.1
Google
Image generation and editing model in the Gemini 3 family, shipping across Gemini, AI Studio, the API, Flow and Stitch, at about half the API price of Nano Banana 2.
Official source ↗Comfy Agent
Comfy Org
An agent inside Comfy Cloud that plans, builds, runs and fixes ComfyUI workflows on your canvas. Now available to everyone in Comfy Cloud.
Official source ↗OpenAI discontinued the Sora app on Apr 26, 2026 and the Sora API on Sep 24, 2026, so it is not listed.