Re-running an old castle prompt: Fable 5.2's procedural Blender output is a visible jump
aitrendz_xyz · x · 2026-11-10
Re-running the same September procedural Blender castle prompt on both models: Fable 5.2 took 71 min, 380k output tokens, $32.9 in API terms; Opus 5.5 took 55 min, 324k tokens, $19.1. Fable's detail density and cleanliness mark a clear jump, per the author.
More from coding & agent
- Fable 5.2 built an interactive 3D globe of Earth's history from one prompt, ~$550 in tokens — Chris_Brannigan · 2026-11-10
- Adobe ships MCP support across its entire Creative Cloud stack, from Photoshop to Firefly — LinusEkenstam · 2026-11-10
- Hugging Face Transformers merges decision models primitive: Jev, KEV, CLEF, Decider — multimodalart · 2026-11-10
- Microsoft's LongCapHarness boosts Llama3-8B long-context by 64 points with zero training — omarsar0 · 2026-11-10
- One message to a realistic UGC video: Fish Audio's Creative Agent workflow — socialwithaayan · 2026-11-10
- Dot quietly ships multi-exec, letting your agent drive your computer and browser — paw_lean · 2026-11-10