Claude Fable 5.2: what it changes for your images and videos
Anthropic released Claude Fable 5.2 on November 9: faster, cheaper, better at vision. What it changes if you create images and video from Claude.
Frank HoubreFounder of Imaginode
4 min read
Fable 5.2 is out, and it costs less
On November 9, 2026, Anthropic released Claude Fable 5.2 and Claude Mythos 5.2, its most capable models. They share the same underlying model: Fable is the version open to everyone, Mythos is reserved for vetted organizations.
According to Anthropic's announcement, both models improve in coding, reasoning and vision. They generate outputs up to 30% faster than version 5.1, for $8 per million input tokens and $40 per million output tokens. An optional fast mode speeds up generation by up to 2.5 times, at twice the price.
Fable 5.2 is available on all platforms, including Amazon Web Services, Google Cloud and Microsoft Azure. Mythos 5.2 stays reserved for cybersecurity and biomedical research teams that pass Anthropic's verification programs.
Anthropic also gives a telling example: starting from an empty folder, Fable 5.2 coordinated over 1,000 subagents for eight days to write an operating system that boots on a laptop.
The numbers that matter to a creator
Most of the published benchmarks are about code and science. Two points still touch visual work: vision, and the cost of a task.
On Chartography, a test of reading charts and figures, Anthropic reports 66.4% for Fable 5.2 at medium effort, against 64.4% for Opus 5.5 at max effort, for less than a tenth of the cost per task. It is not a test of creative images, but it points to a model that reads what you show it better.
Anthropic also stresses cost: at high effort, Fable 5.2 matches Opus 5.5's best score on a coding test for roughly half the cost per task. In other words, a stronger model is no longer necessarily a more expensive one to run.
And Anthropic itself says Fable is not for everything. For most tasks it recommends starting with Claude Opus 5.5, and moving up to Fable only when the work gets long, ambiguous or highly technical.
Claude Fable 5.2: what Anthropic announced on November 9, 2026
| Point | What Anthropic announced | For you |
|---|---|---|
| Price | $8 per million input tokens, $40 output | A long project driven by Claude costs less than with 5.1 |
| Speed | Up to 30% faster than 5.1, fast mode up to 2.5 times | Less waiting between steps of a project |
| Vision | 66.4% on Chartography, ahead of Opus 5.5 | Better at reviewing a storyboard or a page |
| Pro subscribers | Fable up to 50% of weekly limits, one to three resets | Enough to test on a real project this week |
| Recommended use | Opus 5.5 first, Fable for the most demanding work | No need for Fable on a single image |
What it changes if you create from Claude
Claude does not generate images or videos itself. It does so by calling a tool, for example Imaginode through MCP. So the model does not change the quality of a render, it changes the quality of the direction.
For a single image, the difference will be small: the image generator does the work, and a lighter model writes a prompt very well. No need to burn your Fable quota on a thumbnail.
Fable 5.2 should show on a whole project: cutting a three-minute video into thirty shots, writing one prompt per shot while keeping the same characters, launching the renders, reviewing the results and redoing the ones that drift. That is exactly the kind of long, ambiguous task Anthropic positions the model for. We have not measured it ourselves yet.
If you have not connected Claude to a generator yet, our guide to generating images and videos from Claude takes two minutes. And to estimate what the renders themselves cost, still the biggest part of the bill, see what an AI image or video costs.
What I would do this week
Use this week's extra Fable usage to test on a real project, not on a single image.
Pick a project with several shots: a music video, an ad, a short sequence. Give Claude the script and your character references, and ask it to propose the shot breakdown before generating anything.
Approve the breakdown, then let it launch the images one by one and review each result. Count how many renders you had to redo: that is the only number that will tell you whether Fable is worth its quota for you.
On the Imaginode canvas, every render launched from Claude arrives with its prompt, next to the other shots. You see at a glance what holds and what drifts.
How we analysed this
Written on November 10, 2026 from Anthropic's November 9, 2026 announcement, read on November 10. The screenshots come from that page, captured the same day. The speed, price and benchmark figures are Anthropic's, not independent measurements.
We have not yet measured Fable 5.2 on image or video projects: what concerns visual creation is our analysis, not a test. Disclosure: Imaginode, Frank's product, is a tool for creating AI images and video, usable from Claude through MCP.