After the new model was released, developer Riley Brown posted a 3D short film titled “The Edge” on X.
https://x.com/rileybrown/status/2119902109709537708
The scene unfolds around a work desk. A MacBook on the desk lights up, Claude’s lobster mascot Clawd moves around the scene, and keyboard presses, screen content, and ambient lighting work together. The camera cuts back and forth as the sky outside changes from dusk to night, finally giving way to sunrise.
The short film contains 13 shots and runs for about 59 seconds.
Riley said he gave Fable 5.2 only a single prompt during the livestream. The model then wrote its own Blender code to create models of a desk, laptop, lights, and a character, choreograph camera movements, and animate the objects.
After completing the modeling and animation, Fable used 10 GPUs on RunPod for 4K rendering, a process that took 52 minutes. Riley reported that the cloud rendering cost $5.75. The model also color-graded the footage, created the titles, and used ElevenLabs to generate the soundtrack and sound effects.
Another developer, Peter Yang, used Fable 5.2 to create a 20-second autumn travel advertisement for Fushimi Inari Taisha in Kyoto, Japan. The entire video, including the music, was built through code.
https://x.com/petergyang/status/2119886560992854431
In fact, as early as September, Opus-generated videos had already gone viral.
On September 14, developer Kevin Ngo showcased an animation drawn frame by frame in JavaScript by Claude Opus 5.
https://x.com/kevin_t_ngo/status/2099477219877978289
Developers then used Claude Code to recreate this Risograph screen-printing style, generating roughly 3,500 lines of Canvas 2D code that implemented 26 illustrated scenes in a single HTML file and synthesized the music with WebAudio.
On September 22, Anthropic released Opus 5.5, further fueling this wave of creativity.
On the day of the release, Reddit user AzorAhai1TK shared a surreal horror short of about one minute. The brief he gave Claude was deliberately open-ended: create a video depicting an ordinary workday with an eerie atmosphere, with both the visuals and audio produced programmatically and the specific tools left to the model’s discretion.
https://www.reddit.com/r/ClaudeAI/comments/1wnh4fn/opus_55_one_shot_video_generation/
According to the creator, Opus 5.5 spent 45 minutes in Extra-High reasoning mode completing the production, including the final render. The toolchain included NumPy, SciPy, Cairo, Piper TTS, and FFmpeg.
More ambitious works began appearing over the following days.
On September 24, a Reddit user showcased a combat video in the style of a retro SNES game. Opus 5.5 orchestrated multiple agents, each responsible for characters, scenery, music, and combat animations. The author said no pre-made assets were provided, but later confirmed in the comments that the production went through two or three rounds of revisions.
https://www.reddit.com/r/singularity/comments/1worlfs/opus_55_is_insane_at_making_videos/
Around the same time, X user @IterIntellectus posted a 2-minute, 16-second animated history of Western civilization, using black-and-gold lines and dynamic graphics to connect historical periods such as ancient Greece, Rome, the Renaissance, and the Industrial Revolution.
https://x.com/IterIntellectus/status/2103212539895017864
The video received 5.8 million views roughly 24 hours after being posted and later surpassed 15 million total views. Its creator said it had taken only one prompt.
The scale of this wave of enthusiasm was also considerable. The third-party video index Claude Video currently catalogs 1,276 videos created with Opus 5.5 by 1,007 creators, 457 of which include public prompts. According to the site, they have generated approximately 78.9 million cumulative views on X.
https://claudevideo.org/
How exactly does Claude generate videos?
Claude can write HTML, CSS, JavaScript, SVG, or Canvas animations based on natural-language requirements, use Three.js to build 3D scenes, or invoke Blender for modeling and rendering.
For example, when asked to create a 15-second product promotional video, Claude can first plan the shots and timeline, then write code controlling text, graphics, camera movements, and transitions. The program calculates each frame based on time, before the final MP4 is exported through browser recording, a rendering engine, or FFmpeg.
When complex 3D scenes are required, Claude can also write Blender Python scripts to automatically create models, assign materials, position lights and cameras, and then invoke the GPU to perform the rendering.
This process relies primarily on the model's capabilities in programming, spatial reasoning, task planning, and tool use. The visuals are produced through programmatic computation and rendering, creating a clear technical difference from the generative approach used by video models such as Veo and Runway.
Code-generated video has advantages of its own.
The positions of text, numbers, and graphics can be controlled precisely; scenes and animation timelines can be revised repeatedly; and interface elements in product demonstrations, technical diagrams, and animated charts can all be generated using structured code.
Earlier, Remotion also released a Claude Code plugin that enables developers to create and edit video projects through natural language. On October 8, Anthropic officially launched the Claude Motion beta, which supports turning text, charts, and graphics into animations and exporting them as MP4 files. It is available to Team and Enterprise users.
At present, the most mature applications of Claude's code-based video generation are concentrated in motion graphics, information visualization, product showcases, technical explainers, and stylized 3D animation. For highly photorealistic people, complex natural movement, and continuous cinematic shots, professional generative video models still retain their respective advantages.After the release of Fable 5.2, its robotic manipulation capabilities also sparked discussions among several Anthropic researchers. On social media, they said that the robotics field’s “bitter lesson” may have arrived.
From Opus 5.5 in September to Fable 5.2 today, Claude’s video generation has continued to attract attention. Will dedicated video generation models also face their own “bitter lesson” in the future? What do you think?
