Sandor Dobi

@norfmenn @0xSero Droid -> /settings -> compaction limit overrides. you can setup model by model the limits. If you are BYOK don't forget change the compaction model too. Because the default is Sonnet

@Dobisoft @0xSero How did you set up auto-compression?

Droid - GLM-5.1 I asked it to install Ghidra Pointed at compiled typescript for a popular product "Reverse engineer this and provide me code that compiles to it then make it so I can use the app from my own dev build" 20 million tokens 1 shot later Full application, compiles down to the same thing, usable, can rip out any BS Making a bet, ZAI will have the first non-western #1 model in almost all benchmarks soon. Don't discount the value of training data, people who use ZAI are especially from the west have a higher technical cieling. Model is excellent. 432$ a quarter well spent.

@0xJaatster @shivam_nerd I using BYOK from opencode. It works perfectly, using cache and all

@shivam_nerd any reason why the cost is still 0.55x even though the model is much cheaper? does Droid Core have any specialisation done to make the model better that explains the cost?

I think I am probably the biggest user of the GLM models at Factory. GLM-5.1 is a major improvement - super fast and significantly better. It is amazing as a daily driver, and super cost-effective as a worker model in missions. Try it out and let us know what you think!

@shivam_nerd I too using only glm-5.1 in droid nowadays

@0xSero Do you forgott to cancel in time?

I regret to inform you I have paid for Claude again. 🤦

@NoeFlandre @0xSero Max 10 mins, far from an hour.....

@0xSero Ping us back in an hour when u run out of tokens :)

Horváth András remekel a héten. 😆

@robinebers That is why I banned all Anthropic garbage. GPT and GLM in tandem the way

built all day with Opus 4.6 now had GPT 5.4 High Fast review it guess what it said

@meltingdiodes @0xSero @lumendriada @bytebot @Everlier I tried a lot (Claude Code, opencode, droid) with plugins (superpowers, oh-my-openagent) and without. Droid gives me the most stable result with the most minimal trouble.

@0xSero @lumendriada @bytebot @Everlier why droid over cmux and opencode, 0x? is their cli good?

Thank you for baring with me while I travelled, finally got a minute to wrap this up! Winners for 100M Factory tokens announced in the second post I hope Droid treats you well, it's been an indispensable part of my stack and really has it's own unique flavour of agentic power

@0xSero @lumendriada @bytebot @Everlier Gratz for the Winners!

- @lumendriada - @bytebot - @Everlier I will DM you guys and we can work the details out.

@dealignai and what you propose for a 128GB M4Max MacBook Pro?

MiniMax m2.5 JANG_4M CRACK - production use uncensored: 92% on MMLU, 200 questions 92% on harmbench 120gb 4bit average 50 token/s imo, minimax is king o 2025-2026. theres no better model in terms of balance of speed, knowledge, and size. huggingface.co/dealignai/Mini…

4/ AI-Native Visualization Droids decide how to present their work. They render Mermaid diagrams, charts, tables, dashboards, and metrics directly in the conversation. The UI is dynamic and generated by Droid.

3/ Multi-Agent Sessions Run Droids across multiple projects at once. The sidebar lets you switch between active sessions, monitor progress, and jump back into any Droid's work without losing context.

2/ Computer Use Droids in the desktop app control other applications. They navigate VS Code, interact with browser tabs, read documents, open excel, and work with whatever is running on your desktop.

1/ Droid Computers Every Droid gets a persistent machine it can call home. Register a Droid Computer in the cloud, or on your device at home. Access it from anywhere easily via the web, desktop, or mobile.

Today we're releasing the Factory desktop app. A native interface for autonomous AI agents that work across every part of your software business.

@0xSero M5 Ultra is nice. I’ll buy it too IF only will have 512GB RAM option

Framework Desktop - arrived 1x RTX 6000 - arriving this week 3x RTX 6000 - purchasing in a few days 8x RTX 3090 - donating to a friends lab Intel cards - being donated to me soon Mac M5 Ultra - purchasing at launch 1TB of memory very soon

@0xSero Two theories (both of mine). 1. This is only PR stunt to fade the OSS advances 2. This is the start of weaponizing AI and end of democratizing it But Chinese models will win just because of sheer volume and capability to outrun the frontiers.

“Hey everyone. We built a model that can pawn all of you, but it’s so dangerous that we won’t release it. Why did we make it? Don’t ask questions silly, you’re safe in our capable hands! We definitely won’t hurt you! We just want to protect you. We know what’s best for you”

Én elmegyek szavazni vasárnap. Te is jössz? cselekedj.magyartisza.hu/szavazok

@itsolelehmann If MacBook Pro the max ram (128GB). I have it and it can fit some nice local models like Qwen3 next coder. I am waiting for a new Mac Studio later this year but only then if they will sell it with 512GB RAM again. Unfortunately this may take time, maybe I'll wait for it a year.

is it worth it to get the m5max macbook pro? I want to be able to run local models. how much ram should I get? can't get a mac studio because I'm traveling a lot this year.