Weekly digest
The week on HuiaHub: sandboxed agents, RAG tooling and local models
41 new HuiaHub pages this week: sandboxing coding agents, RAG retrieval, local models on Macs and phones, Claude Haiku 5.5 and six how-to guides.
4 min read
Signals
Weekly digest
What changed this week across the tools, projects and skills we track. Every item links to its source.
Written by HuiaHub from primary sources. How it's made
NewsHow NVIDIA fine-tuned Nemotron 3 for gold-level IOI and IMO 2026 results
NVIDIA details how SFT, RL and test-time loops took Nemotron 3 to gold-level IOI and IMO 2026 results, and which checkpoints and datasets are open.
4 min read
NewsMicrosoft rebuilds Agent Lightning to train agents in their real harnesses
Microsoft rebuilt Agent Lightning as a 3,500-line RL framework that trains agents through the harnesses they already run. What changed for RL teams.
4 min read
- News
GitHub Copilot local sandboxing reaches general availability
GitHub made local sandboxing for Copilot generally available, limiting what agent-run commands can touch. Setup, OS needs and gaps for teams.
4 min read
- News
GitHub moves AI secret detection to a fine-tuned model and AI Credits
GitHub moved AI secret detection to a fine-tuned model and set AI Credit billing for push protection and Copilot checks. What changes for security teams.
4 min read
- News
Claude Haiku 5.5 brings lower prices and an effort setting to Haiku
Anthropic's Claude Haiku 5.5 cuts Haiku input pricing to $0.10 per million tokens and adds effort control. What API teams must change to migrate.
4 min read
NewsModels & inferenceIntel details the GPU D270, an air-cooled inference card with up to 480 GB
Intel detailed the GPU D270 inference card: 480 GB LPDDR5X on partner cards, 160 GB on its own, 350 W PCIe. No price or ship date yet for buyers.
4 min read
NewsAssistants & chatCharacter.ai enters Chapter 11 and puts the company up for a court-supervised sale
Character.ai began a Chapter 11 sale process on October 12, 2026. What stays the same for users for now, and how to export chats before the sale ends.
3 min read
- NewsCoding & dev tools
QGIS bans LLM, AI and agent-assisted code in its main repository
QGIS approved a policy that bans LLM, AI and agent-assisted code in qgis/qgis. What changed for contributors, how the vote went and what is still open.
5 min read
- NewsModels & inference
Gemini 4 Argon reaches the Gemini API, AI Studio and the Gemini app
Gemini 4 Argon is now on the paid Gemini API at $2 input and $10 output per 1M tokens until November 30, with a 1M-token window and long decode.
4 min read
- NewsModels & inference
NVIDIA releases Nemotron 3.5 Super VL and a 24GB Bolt model for local agents
NVIDIA released Nemotron 3.5 Super VL, a 120B open vision model, and Bolt 74B-A9B for 24GB local use. What developers can run now and what is missing.
4 min read
- NewsCoding & dev tools
GitHub AI Scan now checks PHP, Terraform and Dockerfiles in pull requests
GitHub's AI Scan for pull requests is now GA. It flags vulnerabilities in PHP, Bash, Terraform and other code CodeQL skips, billed in AI credits.
4 min read
- News
Google DeepMind releases SynthID Bio code, weights and Nature paper
Google DeepMind published SynthID Bio, with Apache 2.0 code, a Nature paper and model weights for marking AI-generated protein data. What shipped.
4 min read
- News
Google announces Gemini 4 Argon, starting with vetted cyber defenders
Google announced Gemini 4 Argon with a 1M-token output limit and set API prices. Only vetted cyber defenders get it for now; developers must wait.
4 min read
- News
GitHub Copilot can now operate desktop apps on macOS and Windows
GitHub Copilot CLI and the Copilot app can now operate desktop apps on macOS and Windows in public preview. How to turn it on, approvals and risks.
4 min read
- News
Cloudflare releases Clef decision models and a managed RL fine-tuning service
Cloudflare released Clef and Clef-flash, decision models that return typed answers with probabilities, on Workers AI and Hugging Face. Prices, specs and gaps.
4 min read
- News
Aleph Alpha releases Kolibri, an open-weight German and English MoE model
Aleph Alpha released Kolibri, a 78B MoE model with 3.46B active parameters and Apache 2.0 weights. What changed, what it needs to run, and what is unknown.
4 min read
- News
NVIDIA adds a 64GB DGX Spark for local AI, with two-unit clustering
NVIDIA's DGX Spark gets a 64GB configuration at $4,999 from six partners, plus two-unit clustering. What changes for local AI developers.
4 min read
- News
Ai2 opens AstaBrief 8B, the model that writes Asta's fast cited reports
Ai2 released AstaBrief 8B, the open model behind Asta's Fast mode, with its SFT and DPO data. What it does, how it was trained and how to run it yourself.
4 min read
- News
GitHub Copilot deprecates four models: what replaces each one and where
GitHub removed Gemini 3.5 and 3.6 Flash, Kimi K2.7 Code and Claude Opus 4.7 from Copilot on October 2, 2026. Replacements, plan limits and admin steps.
4 min read
- News
Copilot code review gets API requests and a Balanced default effort level
GitHub Copilot code review can now be requested through the REST and GraphQL APIs with a per-request effort level, and Balanced is the default effort.
4 min read
- Weekly digest
This week on HuiaHub: local models, MCP guides and 18 new videos
Local LLMs with Ollama and vLLM, MCP servers in Python, Claude Code hooks, 18 talks and new agent skills: everything HuiaHub published so far.
4 min read