What is mattwood.fyi?

A live, structured feed of riffs and links from Matt Wood, Chief AI & Technology Officer at AWS — readable by people and queryable by agents as a typed knowledge graph.

Is there an API and is it free?

Yes. A public, read-only query API is available with no authentication and no API keys. See https://mattwood.fyi/openapi.json and https://mattwood.fyi/docs.

How should an AI agent use this site?

Fetch https://mattwood.fyi/feed.json for the full structured feed, or query the API for keyword/semantic search and the connection graph. See https://mattwood.fyi/agents/ for when-to-use guidance.

Teaching Qwen to Paint with Code

Connections

Supports

Nova3D: Code-Native Generation of Programmable 3D Assets Both explore code-native generation of visual/programmatic outputs; Nova3D generates 3D assets via code while this item trains models to paint through code execution
Ornith-1.5: Self-Scaffolding to Self-Improvement Both apply self-improvement/RL-style training loops to enhance model capabilities beyond base supervised learning

Related to

Qwen3.8 27B Model Analysis Both focus on Qwen model capabilities and analysis, with the new item specifically exploring RL-based training of Qwen for visual code generation
Qwen 3.8 2.4T Mixture of Experts Model Both concern Qwen model variants; the new item trains Qwen with RL while this covers the Qwen 3.8 MoE architecture
Canon: A Text-Based Game Creator for Kids Both involve generating creative visual/interactive outputs through code, though Canon targets kids and uses text-based game creation rather than RL-trained painting
Neuronpedia Jacobian Lens: Interactive J-Space Explorer for Qwen3.6-27B Both involve deep analysis of Qwen model internals — one probing Jacobian activations of Qwen3.6-27B, the other training Qwen's behavior via reinforcement learning

Runway Solaris: Interface World Model OS Teaching Qwen to paint with code uses AI to generate visual output programmatically; Solaris similarly generates visual interfaces but bypasses code entirely, representing an adjacent but competing approach to AI-driven visual generation
Muse Spark AI Model Both involve teaching AI models to generate visual content through learned representations — Muse Spark via synthesis, Qwen via code-based painting
Introducing OUI-1: Generative UI Model Both involve finetuning models to generate visual/interface artifacts through code-like languages — Qwen painting via code and OUI-1 generating UIs via openui-lang share the code-as-output-medium paradigm
Qwen3.8 27B Quantization Benchmarks Both items focus on Qwen model capabilities and applications, with the quantization benchmarks informing practical deployment decisions for models like those used in teaching Qwen to paint with code
Introducing ChatGPT Images 2.5 Teaching Qwen to Paint with Code explores AI image generation through code, representing an alternative approach to the same domain of AI-driven image creation that ChatGPT Images 2.5 advances

Supported by

Orbify Demo 2 - v72 Both demonstrate creative use of code to produce visually compelling rendered outputs - Qwen painting with code and Orbify's PlayCanvas-powered 3D demo both treat rendering engines as expressive mediums

Type:	link
First published:	Tuesday, 25 August 2026 at 22:58 UTC
Location:	https://surya.website/rling-qwen-to-paint-with-code
URL:	https://mattwood.fyi/i/a9247541

I'm Matt Wood, and this is For Your Information. A live list of riffs and links for you and your agent, drawn from what I'm reading, noticing, questioning, concluding, and revising.

Connections