LLM-first music: agents wrote the score for a full song for about $50 in tokens
Ok-Bullfrog-3052 · reddit · 2026-11-18
The author demonstrates his 2024 prediction: instead of prompting Suno directly, he had "Claude Fable 5.2" and "GPT-6 Astra" agents co-write an actual score in a shared workspace ($50 in tokens over a week), used Gemini 3.8 Flash for feedback, then had a traditional music model "cover" the result. He specified B minor, chorus dissonance, and bandoneon; rendering was optimized by OpenAI models to run on a 4090 in 9 seconds instead of 44 minutes. He argues prompt-driven models like Suno have hit their compositional ceiling and general intelligence is the future of music.
More from Multimodal
- TensorScale Visualizes Every Inference Stage and Benchmarks Against a Leading Hosting Provider — TerryTangYuan · 2026-11-18
- HeyGen launches Avatar 6: 84% of viewers couldn't tell it from a real human — HeyGen · 2026-11-18
- Redditor builds audioreactive NYC subway car with TouchDesigner, UE and ComfyUI, textures via Krea 2 — tekprodfx16 · 2026-11-18
- HeyGen Avatar 6 arrives in ComfyUI with one-prompt direction of lip sync and gestures — HeyGen · 2026-11-18
- YC demo builds a movie inside a meeting: speech-to-video agent streams in real time — ycombinator · 2026-11-18
- Headless Motion: motion graphics as JSON, written and frame-checked by agents — RichardsonDx · 2026-11-18