Product turntables
A bottle, a sneaker or a phone turning on a pedestal, with studio light sweeps and reflections. Clean, brand-safe, no warped logos.
Describe an animation. Anthropic's newest model writes a 3D scene and its sound design in code, your browser renders it frame by frame, and you get an MP4 on your timeline, ready to edit.
Claude Fable 5.2 is the publicly available version of Anthropic's newest frontier model, released on November 9, 2026 alongside the restricted Claude Mythos 5.2. Anthropic positions it for large coding projects, long agentic work, reasoning and vision, and says it creates simulations, animations and educational videos with sound in code. That last part is exactly what the Claude model of Nano Banana Video Studio does.
Status as of November 10, 2026. Sources: Anthropic, CNBC, Fortune.
No render farm and nothing to install: Claude Fable 5.2 writes the scene, your browser plays it back frame by frame and encodes the video, and the clip lands on your timeline.
Pick "Claude Fable 5.2" in the Video Studio, write what should happen, choose a format and a duration from 3 to 60 seconds.
It writes a three.js scene: models, materials, lights, camera moves, timing, plus a matching sound design built with Web Audio.
Your browser renders every frame in 720p or 1080p and encodes an H.264 MP4 with AAC sound. A scene that crashes is sent back to Claude to fix.
Ask for a change in plain words ("slower camera", "warmer light"), or edit the code yourself. Then mix it with AI video clips on the timeline.
Code-made animation is exact: every move is timed to the frame, text stays sharp and loops are seamless. That is where Claude Fable 5.2 beats a video model.
A bottle, a sneaker or a phone turning on a pedestal, with studio light sweeps and reflections. Clean, brand-safe, no warped logos.
Particles that assemble into your logo, kinetic type, a sound hit on the reveal. Exact timing for intros and outros.
A rotating solar system, a beating heart, a data pipeline drawn step by step. Anthropic itself highlights educational videos with sound.
Bars that grow, maps that light up, numbers counting to a total. Give the figures in the prompt, they stay correct on screen.
Seamless abstract motion for stream overlays, website heroes, event screens and music visuals.
A 3D shot between two AI video clips: put a Fable 5.2 packshot after a MiniMax H3 lifestyle scene on the same timeline.
Name the subject, the camera, the light, the timing and the sound. Replace the parts in [brackets], paste into the Video Studio with Claude Fable 5.2 selected.
A [matte black water bottle] on a round pedestal, turning one full revolution over the whole video. Dark studio, a soft key light from the left, a thin light sweep crossing the bottle every [3] seconds. Slow push-in from a medium shot to a close-up. Sound: a low ambient pad and a soft shimmer on each light sweep.
Hundreds of [golden] particles drifting in darkness, then pulled together into the word "[BRAND]" in a bold sans-serif at [2] seconds. Hold, then a subtle glow pulse. Camera static, slight depth of field. Sound: rising whoosh, a deep impact when the word forms, a short sparkle tail.
[The solar system]: the sun in the center, [8] planets on their orbits at relative speeds, labels fading in next to each planet one after the other. Camera slowly orbiting from above to a 30-degree angle. Calm educational tone. Sound: a soft space ambience and a light chime as each label appears.
A clean bar chart on a white background: [5] bars for [2022 to 2026] growing one after the other to [12, 18, 25, 41, 63], the value counting up above each bar. The last bar highlighted in [yellow]. Camera static. Sound: a soft tick per bar and a satisfying chime on the last one.
An abstract field of [glossy spheres] gently floating and colliding in slow motion, pastel [pink and blue] light, reflections everywhere. The last frame must match the first so the video loops seamlessly. Sound: an ambient drone with no clear start or end.
The words "[Summer sale]" appearing letter by letter with a bouncy overshoot, then "[up to 50% off]" sliding in below. Bright [orange] background with soft moving shapes. Ends on a 1-second hold. Sound: a playful pop per letter and a whoosh for the second line.
Three ways to make a clip in the same studio. Prices are for a 5-second clip in 1080p.
Scores published by Anthropic at launch, with safeguards on and at the highest effort. Company-reported results, not independent tests.
Source: Anthropic, Claude Fable 5.2 page. GPT-6 Astra pricing: Fortune. Full breakdown in our Claude Fable 5.2 guide.
Use the Claude API with claude-fable-5-2 to plan a campaign and write one prompt per shot, then send each shot to the Nano Banana REST API or MCP server and get a video URL back.
import Anthropic from '@anthropic-ai/sdk'; // 1. Claude Fable 5.2 writes the shot prompt const claude = new Anthropic(); const msg = await claude.messages.create({ model: 'claude-fable-5-2', max_tokens: 1024, messages: [{ role: 'user', content: 'Write one 9:16 video prompt, under 60 words, for: ' + brief }] }); // 2. Nano Banana renders it const res = await fetch('https://nanobananavideo.com/api/v1/text-to-video.php', { method: 'POST', headers: { Authorization: `Bearer ${process.env.NB_API_KEY}`, 'Content-Type': 'application/json' }, body: JSON.stringify({ prompt: msg.content[0].text, video_model: 'veo3', duration: 5 }) }); const { video_url } = await res.json();
Something else? Contact us. Using Claude as an agent? See Claude AI video generator.
Describe it below. Claude Fable 5.2 writes the scene and the sound, Nano Banana renders the MP4.
Building a pipeline? Explore the API