🔍 Review Mode: One-Click Fix Suggestions

Review mode now surfaces one-click suggestions to apply fixes directly. Based on what the reviewer finds, it'll recommend switching to Code, Debug, Architect, or Orchestrator mode—go from "here's the problem" to "fix it" without extra navigation.

🧠 Claude Sonnet 4.6

Now supported across Anthropic, Bedrock, and Vertex providers. Free for a week in Kilo for Slack Code Reviewer.

🤖 New Models & Providers

Providers: Apertis, Zenmux
Models: MiniMax M2.5, MiniMax M2.5-highspeed, MiniMax M2.1-highspeed, GLM 5 (now default for Z.ai), MiniMax 2.1 and GLM 4.7 on Fireworks.ai
Embedders: Voyage AI support, custom embed dimensions for Ollama

✨ Quality of Life

Agent Manager scroll fix, improved user message contrast with theme-aware colors, better formatting for sub-cent token costs, dev container persistence for threads and settings, chars count in ListFiles tool.

Code Reviewer Roast Mode

Brutal, unfiltered PR feedback—the kind a very blunt senior engineer might offer at 4pm on a Friday. "You named this variable temp2. Temp for what? Temporary sanity?"

🏙️ Kilo x Gas Town

Self-hosting Gas Town works, but you're signing up to own every failure mode—persistent compute, key management across providers, and being your own on-call when things break overnight. Kilo x Gas Town handles all of that.

🦞 KiloClaw

OpenClaw (209k+ GitHub stars) is the fastest-growing open source AI agent project right now. It connects to 50+ chat platforms, runs shell commands, controls browsers, manages files, and remembers everything across sessions. KiloClaw hosts it for you—one-click deploy, zero to running in under 60 seconds.

🎙️ The Kilo Show — This Friday

Join us Friday at 20:30 CET / 12:30 PM Eastern / 9:30 AM Pacific. Five attendees win $50 credits + Kilo T-shirts.

📖 Auto Model: Stop Thinking About Which AI Model to Use

You're deep in a coding session, architecting with a powerful reasoning model. Then you switch to implementation and realize you're still burning through tokens for straightforward code generation. The whole dance gets old fast.

This Week's Contributors 🎉

@theQuert – Apertis API provider
@Astricaelus – MiniMax M2.5 models
@nubol23 – Voyage AI embedder support
@CoinAnole – Moonshot coding endpoint fix
@Githubguy132010 – Dev container persistence, Agent Manager scroll fix, user message colors, file deletion checkbox fix
@Patel230 – maxReadFileLine=0 unlimited, symlink handling
@mikij – Agent Behaviour settings visual fixes
@Drilmo – Filter internal verification tags
@wombatepiclandingstudio – Context token indicator flicker fix
@benzntech – MODEL_NO_TOOLS_USED error loop fix
@abdulrahimpds – Settings search bar UI fixes, total API cost persistence
@Madrawn – Slash commands in tool_result blocks
@Leoyzen – Preserve reasoning content in OpenAI format
@rayss868 – Honor explicit 'disable' for reasoning effort
@hdcodedev – Kill Command button reliability, prevent duplicate tool IDs, images lost on edit fix
@SenatusSPQR1 – NanoGPT native reasoning field extraction
@shssoichiro – Custom embed dimensions for Ollama
@evanjacobson – Prevent MCP server restarts, fix provider settings leak, prevent abort listener memory leak
@dacsang97 – Anthropic extended thinking compatibility
@Schrolli91 – Better formatting for low cost values
@Neonsy – ZenMux context window detection, GLM 5 support and Z.ai default model update
@hsp-sz – Zenmux provider
@otterDeveloper – MiniMax 2.1, GLM 4.7, and model updates for Fireworks.ai
@YuriNachos – Disable zsh history expansion
@naga-k – Prevent thinkingLevel being sent to unsupporting Gemini models
@CaiDingxian – Chars count in ListFiles tool

Want to contribute? Check out the repo →

Twitter · Discord · GitHub

© 2026 Kilo Code Inc.
Unsubscribe · Manage Preferences