Everyone’s Testing Claude Fable 5.1 On Code. It Made Me A 37-Second Film.

Summary
Nate B. Jones reviews Anthropic’s Claude Fable 5.1 across complex knowledge-work tasks, comparing its outputs against Claude Fable 5 and OpenAI’s GPT-5.6 Sol. He evaluates how effort settings affect financial modeling and slide generation, tests concise explanatory writing, examines its token pricing, and demonstrates an architectural walkthrough film generated purely from Python code in Blender.

What is shown

• [00:01] Clip of a 37-second 3D architectural animation of a house generated in Blender by Fable 5.1 from a single Seattle property address.
• [00:36] Fable 5.1 at “Low” effort: output of a 7-sheet financial workbook and 13-slide presentation evaluating an acquisition of GoPro by Starman.
• [04:13] Overview comparing four runs on the M&A valuation assignment: GPT-5.6 Sol (Extra High), Fable 5 (Extra), Fable 5.1 (Low), and Fable 5.1 (Extra).
• [06:06] Fable 5.1 at “Extra” effort: 9-sheet workbook and 15-slide deck featuring uncertainty modeling (85% close probability, $0.45 break value), WACC calculation, and 26 linked sources.
• [07:18] GPT-5.6 Sol’s run at Extra High effort: 10-sheet workbook and 10-slide deck with dedicated, verifiable sources and formula-check sheets.
• [09:20] A 100-word writing prompt explaining Toyota’s entry and rise in the US auto market, comparing the drafting styles and causal clarity of Fable 5, Fable 5.1, and GPT-5.6 Sol.
• [12:57] Extended side-by-side demonstration and critique of the 3D Blender walkthroughs produced by Fable 5.1 (37.0s), Fable 5 (35.5s), and GPT-5.6 Sol (12.0s).
• [15:01] Breakdown of API pricing cards and caching rate changes for Claude Fable 5.1.

Claims & numbers

• The presenter notes standard API rates for Claude Fable 5.1 are $10 per 1M input tokens and $50 per 1M output tokens, identical to Fable 5.
• The presenter notes prompt cache read pricing dropped 75%, from $1.00 to $0.25 per 1M tokens.
• Anthropic reports typical workload costs are approximately 25% lower than Fable 5, and highly agentic workloads cost up to 45% less due to caching.
• Claude Fable 5.1 costs twice as much for inputs and outputs as Claude Opus 5.
• Architectural video runtimes generated from code were 37.0 seconds for Fable 5.1, 35.5 seconds for Fable 5, and 12.0 seconds for GPT-5.6 Sol.
• In the M&A model test, Fable 5.1 Low generated 7 sheets and 13 slides with a $1.15 base case; Fable 5.1 Extra generated 9 sheets, 15 slides, 26 linked sources, and an 85% close probability; GPT-5.6 Sol generated 10 sheets and 10 slides with a $1.21 base case.

Notable quotes

• [02:31] “You just don’t need to take the Ferrari to the grocery store. Sometimes, you’re fine taking the Honda.”
• [03:26] “Code will tell a model when it is wrong... Knowledge work does not give you the courtesy of saying I am done.”
• [14:49] “It is where I would go if I were using Blender to communicate a concept in video form.”

Assessment
This is an independent hands-on review and comparison using real outputs from the models rather than cherry-picked marketing demos. The presenter transparently highlights flaws across models, noting missing audit sheets in Fable 5.1 Low and stylized visual shortcomings in the Blender render.

Described by gemini-3.8-flash on 2026-09-29 from the video’s audio and frames.