Anthropic's new flagship Fable 5.5 tested: 2 hours and 40 minutes of continuous pure-code generation produces a 66-second animation

On November 2, X platform AI developer Chetaslua released an animated short film and claimed that it was completed independently by Fable 5.5. According to his introduction, the short film is 66 seconds long, has a resolution of 1920×804, and a frame rate of 24fps. The visuals include foxes, paper birds, clouds, text, and lighting effects, and the background music and sound effects were also generated by Python code written by the model.

This is a modal window.

A network error caused the media download to fail part-way.

Autoplay

Unlike AI tools such as Sora and Runway that directly generate video footage, this demo showcases a technical path for making animations through code. The developer said that no external images, asset libraries, audio samples, or font files were used throughout the process; the model independently completed scene design, image rendering, and sound synthesis, and packaged the final work into an HTML file.

In addition to animation production, several overseas developers also demonstrated interactive 3D scenes reportedly written by Fable 5.5, including ancient sailing ships, ocean waves, and sports cars. The related demos involve front-end graphics technologies such as Three.js, controlling models, lighting, and object movement through programs, and some scenes can be generated with only simple prompts.

In several other comparative tests of complex animations, developer Jazil and others showed the results of different models generating bicycles, animal characters, and character battle scenes. According to the testers, Fable 5.5 was able to complete the coding and running of complex dynamic scenes in about one hour, while some existing Claude models showed problems such as incomplete code and failed animation execution on the same tasks. However, these tests still lack a unified evaluation environment and cannot be directly regarded as a ranking of model performance.

In fact, clues about the release of Fable 5.5 had appeared many times before. In mid-October, an entry named claude-fable-5-5 briefly appeared in the Azure catalog of the third-party model tool LiteLLM and was subsequently removed. Recently, developers also discovered that the text Claude Mythos 5.5 had appeared in the HTML source code of some pages on Anthropic's official website, sparking speculation that a new model was about to be released.

According to Anthropic's existing product system, Fable and Mythos use the same underlying model but implement different levels of safety restrictions. Mythos is mainly aimed at vetted cybersecurity and life science research institutions, while Fable is opened to ordinary users and developers through additional safety protections. However, Anthropic has not yet confirmed whether the 5.5 series continues this design.

Along with the exposure of the new model, a repository claiming to contain Mythos model weights recently appeared on the Hugging Face platform, with a parameter scale reportedly reaching 14 trillion. However, researchers found upon inspection that some of the weight data showed obvious signs of artificial synthesis, and there is no reliable evidence to support the so-called model leak.

This round of model exposure coincides with a critical stage in Anthropic's push toward an IPO. The company's prospectus disclosed on November 2 shows that its revenue in the first three quarters of this year reached $33.242 billion, but continuously growing computing power investment and huge losses have also made the outside world pay more attention to whether the next-generation model can bring actual technological breakthroughs and commercial returns.

As competition among AI models shifts from code completion to long-duration autonomous execution of complex tasks, the animation production and 3D scene generation capabilities demonstrated by Fable 5.5 reflect the potential application space of AI coding agents.

As of press time, Anthropic has not officially confirmed the release time, performance metrics, or availability scope of Fable 5.5, and the above test results still await official confirmation and independent evaluation.