Darko

Made a mini site of how real people name their OpenClaw bots. Check it out: kilo.ai/kiloclaw/names

PinchBench and KiloClaw are dominating

@olearycrew That's pretty crazy traction!

You love to see the PinchBench!

Last week, Kilo Code Reviewer reviewed 525 PRs across 5+ repos in Kilo’s production codebase. Most reviews used Opus 4.6. The average cost per review with Opus 4.6: $1.14. This is 10-100x cheaper than most products out there.

@dorutu_ How much do you get for $10?

opencode go is the best value for money

Kilo Gateway is now part of OpenClaw

Paris. After dark. ✨🇫🇷 We built this with Kilo CLI + MiniMax 2.5. This is an interactive 3D cinematic “Night in Paris” built with Three.js. It's showing the Eiffel Tower, shader-driven Seine, orbiting camera, soft shadows, ACES filmic tone mapping.

Introducing...ROAST MODE. You pushed a PR. You need a review. An honest review. A brutally honest review. Kilo Code Reviewer can now roast your PR. "You named this variable temp2. Temp for what? Temporary sanity?" "I see you copied this from Stack Overflow. The answer had 3 downvotes." Try it now: app.kilo.ai/code-reviews

Kilo Code just overtook OpenClaw to reclaim #1 on the OpenRouter daily leaderboard. 313B tokens processed yesterday; 222B through GLM-5, @Zai_org’s new AI model outperforming Opus 4.5/4.6 across multiple benchmarks. We're also offering free GLM-5 access through Sunday. Get started for free: CLI (OpenCode fork) → kilo.ai/cli (VS Code, Cursor, JetBrains) extensions → kilo.ai/install

@bridgemindai Love your posts! Is there a way to get in touch?

GLM 5 just tied Claude Opus 4.5 on the Artificial Analysis Intelligence Index. The scores: Claude Opus 4.6: 53 GPT 5.2 Codex: 51 Claude Opus 4.5: 50 GLM 5: 50 GLM 5 is matching last generation's best model from Anthropic. One point behind OpenAI's flagship. Now look at pricing: Claude Opus 4.5: $5/$25 per million tokens GLM 5: $1/$3.20 per million tokens GLM 5 delivers Opus 4.5 level intelligence at 1/15th the cost. This benchmark includes GDPval-AA, Terminal-Bench Hard, Humanity's Last Exam, and 7 other evals. Zhipu is not playing games. GLM 5 is the price-to-performance king.

@IATheYoker Hope you like the model!

@TheBinaryNeuron Movimiento muy inteligente

We're making GLM-5 free on Kilo CLI & Kilo Code for a limited time. Learn more: blog.kilo.ai/p/glm-5-free-l…

@RAVIKUMARSAHU78 Yup

@TheBinaryNeuron Huge win for builders, Free GLM-5 on Kilo CLI & Code is a power move.

@Alex_Inspira Agreed!

@TheBinaryNeuron Esto es lo que acelera la adopción 🚀 Cuando un modelo potente como GLM-5 se libera en herramientas prácticas como Kilo CLI y Kilo Code, el impacto en productividad se siente de inmediato.

@0xAlansari openrouter.ai/z-ai/glm-5

@TheBinaryNeuron FP32 or FP8?

@AuliaBismar Do you have a BYOK via the gateway?

@TheBinaryNeuron 429 for both kilo free and my current glm coding plan

@ramanagaadu07 Works for me. What error are you getting?

@TheBinaryNeuron Inaccessible due to huge demand in kilo 🥲

@HighADHD who are you talking about?

@TheBinaryNeuron They're just scammers. Don't fall for their tricks. They only brag about how good their model is compared to the Opus, but they can't even handle the server specs they need. Instead, they decide to rug pull the customers! :))

@LiMzba Love this output! Is there a way to get in touch with you?

Minimax 2.1 is so far the best model I can run locally with good agentic tool use. Wow, such an advancement, can't believe just a year ago, open source models could barely even code a snake game properly. Now I can use a local model to code a self-playing 3D snake game. 2026 will be a very interesting year :)

We brought 120 kilos of crab to @steipete at #ClawCon - please enjoy this cinematic account of how it went down: