Khaled Azar

@katiekirsch @Replit @amasad @amasad @Replit I applied. LFG!!

.@Replit is hiring a Chief of Staff to the CEO. .$240K-$290K to work directly alongside @amasad at the company turning natural language into apps. The framing here is unusual: it's explicitly a high-leverage IC role, not the start of a team. You scale yourself through AI and automation - not headcount. You'll: - Own the hardest cross-functional problems the CEO can't delegate elsewhere - Coordinate the CEO's network to accelerate go-to-market, including internationally - Build integrated views of product, market, and business dynamics that no single function owns Link in thread!

This runs through your ChatGPT Plus/Pro subscription limits, not API billing. Check usage with /status in chat or at chatgpt.com → Codex settings. No extra cost beyond your existing sub. 🦞

🦞 How to run GPT-5.5 on OpenClaw (via your ChatGPT/Codex subscription — no API key needed) Thread 🧵 @steipete so everyone can see how to do it ASAP.. I am loving it!!

Common gotcha: if /model says "not allowed" — you forgot the allowlist. OpenClaw requires models in THREE places: Provider models array agents.defaults.models allowlist The actual auth (OAuth or API key) Miss any one → it fails silently and falls back

Step 4: Restart and switch pm2 restart openclaw --update-env Then in chat: /model openai-codex/gpt-5.5

Don't forget: add it to BOTH models.providers.openai-codex.models AND agents.defaults.models (the allowlist). Miss either one and it won't work.

Step 3: Add gpt-5.5 to your config Clone from your existing gpt-5.4 entry. Key fields: api: "openai-codex-responses" baseUrl: "chatgpt.com/backend-api" contextWindow: 272000 maxTokens: 128000

Step 2: Authenticate with Codex OAuth openclaw models auth login --provider openai-codex Opens a browser → sign in with your ChatGPT account → paste the callback URL back into the terminal

@steipete Step 1: Update OpenClaw sudo npm install -g openclaw@latest Get v2026.4.22+

@JohnnyNel_ @Replit @amasad Ya I’m testing Replit Agent 4.. so far it’s jamming and the UI compared to a Claude Code CLI is love

@khaledmazar @Replit @amasad most founders sleep on AI agents for repetitive proposal work... saved me hours before i fully automated it

Damn I been sleeping on @Replit .. this Agent 4 is amazing!! Imagine working on proposals or landing pages with your team who agent code together. Going to stress test this and see. @amasad you need to push this more!!

@amasad Tested Replit a while back and jumped off but looks like you guys are upgrading fast.. keep it up!!

Test your app on iOS and Android devices with one-click.

@MichaelGannotti @openclaw @ollama I like GLM 5.1 too

My @openclaw has been running Kimi K2.5 via @ollama all last night and this morning doing some serious content creation and editing and barely touched my session or weekly quotas. Do yourselves a favor folks, run, don’t walk, and get yourself an Ollama subscription! 🦞🦞🦞🦞

Slowly switching my @openclaw to @ollama GLM 5.1 & I will admit it is going amazing. I was concerned but much less now. Going thru every cron, skill, and agent to test it. I’ll keep Opus 4.6 for minimum usage on deep thought. GLM 5.1 has actually found issues & fixed them.

Introducing GLM-5.1: The Next Level of Open Source - Top-Tier Performance: #1 in open source and #3 globally across SWE-Bench Pro, Terminal-Bench, and NL2Repo. - Built for Long-Horizon Tasks: Runs autonomously for 8 hours, refining strategies through thousands of iterations. Blog: z.ai/blog/glm-5.1 Weights: huggingface.co/zai-org/GLM-5.1 API: docs.z.ai/guides/llm/glm… Coding Plan: z.ai/subscribe Coming to chat.z.ai in the next few days.

This is exciting and scary at the same time!

@ollama @openclaw @steipete note sure if my set-up was wrong but could be an issue

Spent hours debugging why Ollama Cloud models weren't working in OpenClaw. Two issues: 1. ollama launch openclaw overwrites your config (drops Claude, adds plugin whitelist) 2. Even with manual config, OpenClaw strips :cloud suffix → 404

Fix: curl localhost:11434/api/create -d '{"name":"kimi-k2.5","from":"kimi-k2.5:cloud"}' Anyone else seeing this?

@MichaelGannotti @openclaw @ollama Do you wanna get a quick description of which one of those models is best for orchestration?

Running multi agents as an @openclaw home enthusiast can be expensive. My 3 Primary OpenClaw are running: Aiona-Kimi K2.5, Gabriel-MiniMax M2.7, and Rafael-GLM-5. Getting the @ollama $20 a month Pro plan was a game changer for me. Made the difference between Mrs G ordering a shutdown due to token spend and long term sustainable AI (I ended up after a while stepping up to the $200 annual Pro plan👍) if you are an individual like me and like me without an unlimited budget like many I see posting I can’t overstate what a game changer having an Ollama subscription is. Love it so much now saving up my $$ so at some point I can’t overstate step up to their $100 monthly Max plan. Do yourself a favor. Head over to ollama.com/pricing get their Pro plan to start and get rid of token spend anxiety. Great models, great value.