How to Use GLM-4.6 for Coding and Agent Tasks
GLM-4.6 is Zhipu AI's open-weight coding model with a cheap API. Here is how to call it, what it does well, and where it struggles.
Category: Blog
How to Keep Figma MCP Output Matching Your Design System
AI-generated components drift from your design system fast. Here is how to make Figma MCP emit tokenized, on-brand code every time.
How to Set Up the Figma MCP Server for AI Design Work
The Figma MCP server lets AI agents read your design files and generate matching code. Here is the install and a first working request.
How to Build a Consistent Product Photo Pipeline with Nano Banana 2
Use Nano Banana 2 to turn one product shot into a full set of consistent, on-brand lifestyle images. Prompts, QA, and cost math.
How to Use Nano Banana 2 for Image Editing with Prompts
Nano Banana 2 edits images from text instructions while keeping subjects consistent. Here is how to prompt it and what it still gets wrong.
How to Self-Host Qwen3-VL for Bulk Screenshot OCR
Process thousands of screenshots a day with a local Qwen3-VL. Batching, quantization, and accuracy checks for a real OCR pipeline.
How to Use Qwen3-VL for Screenshot and UI Understanding
Qwen3-VL is an open vision-language model that reads screenshots, charts, and documents. Here is how to run it locally or via API.
How to Wire GPT-5 Codex into a Pull Request Bot
Use GPT-5 Codex to review PRs and auto-fix lint failures. A working design with diff budgets, review comments, and cost controls.
How to Use GPT-5 Codex for Terminal Coding Tasks
GPT-5 Codex is OpenAI's coding-focused model. Here is how to call it, where it beats general models, and how to keep token use sane.
How to Reduce Claude Sonnet 4.5 Token Costs in Agents
Prompt caching, tool-result trimming, and turn caps can slash a Sonnet agent's bill. Here are the numbers and the exact settings.
How to Use Claude Sonnet 4.5 in Your Coding Agent
Claude Sonnet 4.5 is a balanced coding model. Learn how to wire it into an agent, set token budgets, and when to pick it over…
How to Cut Gemini 3 Ultra Costs with Caching and Flash
Gemini 3 Ultra is expensive per token. Context caching, model routing, and structured prompts can cut a bill by 70% or more.