Newsletter Monday 2 November 2026
Daily update about the AI tools we follow. Headlines come from public sources (X, GitHub, Hacker News); the Dutch commentary and summary are machine‑written. Each headline leads to our own news page on that topic; the source link is shown beside it.
In brief
Unsloth has released a 1-bit Dynamic GGUF for Mistral Le Chonk, allowing a 1T‑parameter model to run locally on a Mac Studio. NVIDIA launched Nemotron 3.5 Transcribe, an open 30B MoE model with about 4B active parameters for transcription in 40 languages. Additionally, Evonic v1.6.0 adds support for AI‑agent queries via Telegram and browser notifications. Finally, v0.11 uses a cheaper model to summarize files for Claude Code and Codex, with the caveat of a risk of stale caches.
Codex, Cursor, OpenCode & Hermes
- A cheap model reads my files, so Claude Code and Codex get a summary. Risk: a stale cached one after I change the prompt or provider. v0.11. · source: X · 2026-10-31
A cheap model reads files and provides a summary to Claude Code and Codex, giving indie developers quick code overviews, but a stale cache can occur when the prompt or provider changes.
Repo's, releases & updates
- Unsloth released a 1-bit Dynamic GGUF for Mistral Le Chonk, running a 1T-parameter model locally on a Mac Studio; The Verge dissected Anthro · source: X · 2026-11-01
Unsloth released a 1‑bit Dynamic GGUF for Mistral Le Chonk, allowing a 1 T‑parameter model to run locally on a Mac Studio, giving indie developers access to large models without cloud. - Evonic v1.6.0 released: your agents can ask you anywhere • Telegram questions, tap to answer • Browser alerts when agents need you • Model-f · source: X · 2026-11-01
Evonic v1.6.0 lets agents ask questions anywhere via Telegram and browser notifications, so indie developers can better integrate and respond with their AI agents. - NVIDIA Releases Nemotron 3.5 Transcribe: An Open 30B MoE Model for 40-Language Transcription with ~4B Active Parameters https://www.marktech · source: X · 2026-10-30
NVIDIA launches Nemotron 3.5 Transcribe, an open 30 B MoE model for transcription in 40 languages with roughly 4 B active parameters, providing indie developers with a multilingual transcription tool.