Claude Just Built A Full 3D House In Blender From One Prompt (Fable 5.1)

Summary
Presenter Vaibhav Sisinty evaluates Anthropic's Claude Fable 5.1 model across five complex workflow tests: market research presentation decks, animated SVG graphics, mobile app development, 3D scene creation in Blender, and interactive product websites. Sisinty demonstrates how Claude Fable 5.1 pairs with Model Context Protocol (MCP) integrations to automate end-to-end creative, coding, and spatial tasks from single prompts.

What is shown

• Model Overview & Comparison [02:06]: A breakdown comparing Claude Fable 5.1 and Claude Mythos 5.1 regarding availability, pricing, cache-read discounts, Enterprise Frontier Safeguards (EFS), and safety filtering.
• Benchmark Graph [03:37]: Display of the Terminal-Bench-Science 0.1 benchmark comparing accuracy vs. cost across Fable 5 and Fable 5.1 configurations.
• Test 1: Deep Research Presentation Deck [04:07]: A 12-slide PowerPoint presentation covering 10 AI-native business concepts for 2026, generated after a 20-minute autonomous research run, followed by a full redesign guided by a visual reference image [05:32].
• Test 2: Animated SVG Generation [06:16]: Generation and animation of an SVG depicting a pelican riding a bicycle using raw code, contrasted with outputs from Codex and Gemini [06:53].
• Workflow Integration: OpenArt MCP [07:47]: Demonstrating OpenArt MCP inside Claude to generate financial dashboards for NVIDIA's earnings [08:20], YouTube thumbnails [08:58], and product video storyboard plans ("Smart Shots") [09:34] leading to rendered video clips.
• Test 3: Interactive App Development ("Savor") [12:25]: An iOS calorie tracker app written and running in the iOS Simulator, parsing natural language food logs into visual plates, tracking macros, and providing a calendar view [13:10].
• Test 4: 3D Scene Generation in Blender [14:49]: Using Blender MCP to autonomously build, texture, light, and render a complete modern house environment in Blender, inspected in solid and wireframe modes [15:43].
• Test 5: Apple-Style Product Landing Page [16:17]: A scrollytelling webpage for a Dyson electric toothbrush featuring exploded 3D component animations and spec breakdowns [16:25].
• Prompting Technique & Custom Skill [18:16]: Importing Anthropic's official Claude Fable 5.1 prompting documentation directly into Claude to synthesize a reusable prompting Skill [18:31].

Claims & numbers

• The presenter notes Anthropic released Claude Fable 5.1 alongside Claude Mythos 5.1 on September 1, 2026.
• The presenter states Fable 5.1 is generally available, while Mythos 5.1 is restricted to trusted access programs for sensitive cybersecurity and life sciences work.
• Fable 5.1 costs approximately 25% less than Fable 5 for normal workloads, with cache reads priced 75% lower ($0.25 per million tokens), yielding up to ~45% total cost reduction on multi-step agentic workflows.
• Anthropic introduced Enterprise Frontier Safeguards (EFS) offering zero data retention options for enterprise clients.
• In cybersecurity evaluations, safeguards reportedly reduce false-positive blocks by 60%.
• On the Terminal-Bench-Science 0.1 benchmark shown, Fable 5.1 max scored 52.6% at $37.9 mean cost per task, compared to Fable 5 max at 24.7% at $44.1, while Fable 5.1 low achieved 26.3% at $11.1.
• In the initial research evaluation, Claude spent approximately 20 minutes conducting background research before outputting the final presentation.

Notable quotes

• [01:09] "Every new model comes with its own hidden manual: how to actually talk to it, where it lags, how to save tokens, and what it's genuinely best for."
• [04:24] "You can let the model spend more time working through the task instead of forcing it to answer immediately."
• [14:59] "MCP is basically the bridge that lets Claude talk to Blender. So instead of you manually clicking through every Blender tool, Claude can use that connection to create and change parts of the scene."

Assessment
This is a hands-on review and tutorial showcasing real terminal, simulator, and MCP executions across multiple applications. The demonstrations show genuine working outputs (PowerPoint slides, SVG code, Swift simulator builds, and Blender project viewports), though the generation times are condensed through editing.

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