Pinned
Kilo
4,086 posts
Kilo
@kilocode
Kilo is the all-in-one agentic engineering platform.
3M+ Kilo Coders. Open source since day one!
Replying to @kilocodeThen we handed GLM's winning plan to both models and had each build it from a blank folder. Same 200 users, flag at 35% rollout. Both services turned it on for the exact same 77 users, down to the individual ID. GLM passed all 15 live checks. Kimi passed 14.
The takeaway: once a plan pins down the hard decisions, the model doing the build matters a lot less. The strongest planner writes less and decides more, leaving fewer open questions for whoever builds from it. Full breakdown from @TheBinaryNeuron:- A black hole simulator from a screenshot drop. @MiniMax_AI M3’s vision did the heavy lifting and the whole build ran $0.53.Frontier doesn't only mean Anthropic and OpenAI anymore. I built this black hole simulator by simply dropping an illustration screenshot into Kilo Code, switching to @MiniMax_AI M3, and prompting it to "animate this screenshot into a working black hole simulator". M3's vision
00:00 - For anyone arriving from the cat: our actual read on what the $60B Cursor deal means for model choice, and why we think vertically integrated stacks are the wrong bet.
- Three real sites. Three prompts. $0. This is Step 3.7 Flash, free in Kilo right now, and @coldopn is the one putting it through the work. He's been shipping builds like this constantly lately. Follow him if you want the receipts. P.S. We're also giving away $500 to two peopleWhy aren't more people talking about @StepFun_ai? Their latest model Step 3.7 Flash just made me 3 beautiful websites in less than 10 minutes: 1. Interactive Resume/CV 2. Landing Page for a Car Wash business 3. Custom Wedding Site Total Price: $0 Total # of Prompts: 3 Step
00:00 - EU teams: you can now run top open-weight models with data that never leaves Europe. We've partnered with @inceptron for sovereign, EU-hosted inference in the Kilo Gateway. Kimi K2.6, GLM 5.1, MiniMax M2.5. GDPR and ISO compliant. MiniMax M2.5 starts at $0.15/1M input tokens.
More on what @inceptron brings: > Data residency you can lock to EU-only > ISO-aligned infrastructure > Open-weight models devs are already running daily Full breakdown from @arimesser:
