NVIDIA's Nemotron 3.5 Bolt E4B is an open small language model post-trained from Google's Gemma 4 E4B. It is designed for local voice assistants, game characters and robots, and targets roughly 4GB of video memory on RTX and Jetson hardware.
Moonshot's Kimi K3.1 Pro is now on the Kimi API, with full weights promised by November 4, 2026, and a faster 'Highspeed' tier at twice the price. At $2 per million input tokens and $10 per million output, it costs the same as GPT-6.1 Sol and half as much as Claude Opus 5.5, and Moonshot's own benchmarks put it roughly level with Opus 5.5 and GPT-6 Astra.
Block has announced Dessa, an AI model it says was trained from scratch solely on human conversation. Sample chats shared by the company show exchanges that can feel awkward and slightly uncanny.
In an image of a mountain lake drawn by Claude Fable 5.1, each star encodes a letter according to its height in the sky, a method the model came up with on its own. OpenAI's GPT-6 Astra decoded 44 of the 50 hidden-message images Fable produced.
A coding guide compares Underdog Saluki 27B, a 2-bit version of Qwen3.8-27B, with ISTA's GSQ-RCO quantisation. It looks at GGUF headers, memory use and tool-calling performance on the BFCL benchmark in standard llama.cpp.
The 20th episode of the Interconnects podcast discusses the Kimi K3.1 Pro model, the changing cybersecurity landscape and Muse.