GLM-5.3 autonomous 12-hour run: Prompt for Blender scene generation
majidmanzarpour · x · 2026-09-02
The author released the prompt used to make GLM-5.3-Flash run autonomously for 12 hours to build a Blender scene. Consuming around 100 million tokens, the model started from an empty folder, wrote Python scripts via Blender CLI, and iterated by checking renders until the result met quality standards. The prompt includes detailed task specs, environment constraints, and validation rubrics.
More from coding & agent
- Ontology-driven multi-role LLM loops in production — serendip-ml · 2026-09-02
- Google Cloud Run instances host personal agents for $5.70/month — steren · 2026-09-02
- Agent cuts token usage by 66% and time by half using Minibase tool — jeff_weinstein · 2026-09-02
- Perplexity Computer integrates Coinbase for agent trading — AravSrinivas · 2026-09-02
- Looking for agentic extension that loads whole codebase to context — Round_Ad_5832 · 2026-09-02
- Anthropic's New Model Immediately Supports Native PowerShell Tool Calling — dfinke · 2026-09-02