HaiAI123

Curated Global AI Tools Directory

SurgeIndustry update
↑ 81%

Claude Fable 5.2 gets a 20% price cut, but per-task cost rises at max effort

11/15/2026 — 11/17, 08:19·4 sources (all)·4 reports (all)

Key takeaways

  • Artificial Analysis found Anthropic cut Claude Fable 5.2 token pricing by 20%, yet the same task at max effort cost rose from $12.99 to $26.…

Background & analysis

Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on November 9, local time. According to an IT之家 report on November 16, Fable 5.2 targets complex programming, scientific research, and engineering work, and it also demonstrates an unusual approach to image generation: rather than calling an image generation model, it writes Python code that simulates brushes, paint, and canvas texture stroke by stroke on a virtual canvas. The painting process is broken into many independent brushstroke instructions that the model plans and executes on its own, and it can correct mistakes with new strokes.

The Decoder reported on the evening of November 15 that, in this approach, the model does not compute pixels at all. Instead it places hundreds of thousands of simulated brushstrokes on a virtual canvas, handling bristle texture and paint buildup. Anthropic employee Jake Eaton said a single photorealistic parking garage scene took roughly 700,000 strokes, and he showed the effect on X.

On pricing and cost, testing by Artificial Analysis found that Anthropic lowered the token price of Claude Fable 5.2 by 20%, yet the cost of one task at max effort rose from $12.99 for Claude Fable 5.1 to $26.54, because the new model produces about 1.7 times as many output tokens. The same testing noted that a day later Google shipped Gemini 4 Carbon at a quarter of Fable's price, and that Carbon holds a clear cost edge at max effort. One report argued that the relevant figure is the cost of completing a task rather than the per-token price, and advised teams to benchmark their own workloads across effort settings before switching models.

On November 17, Artificial Analysis posted AA-PersonalityBench results on X, showing that Claude Fable 5.2 is more direct than Claude Fable 5.1 and writes notably shorter responses with almost no em dashes. It ranks higher than Claude Fable 5.1 on Direct and Answer First, but lower on Informativeness and Conviction. Public reporting currently stops at that evaluation, and no further statement from Anthropic on the price and per-task cost gap has appeared.

AI-generated from 4 reports · updated 2 hours ago

Latest turnIn AA-PersonalityBench, Claude Fable 5.2 is more direct than Claude Fable 5.1 and writes notably shorter responses with almost no em dashes. It ranks higher on Direct and Answer First but lower on Informativeness and Conviction.

Related AI tools

updream
24-hour heatHeat index 201 · peak 222 · 3h ago
24 hours agonow

Reports on this story headlines open the original

Today
  1. In AA-PersonalityBench, Claude Fable 5.2 is more direct than Claude Fable 5.1 and writes notably shorter responses with almost no em dashes. It ranks higher on Direct and Answer First but lower on Informativeness and Conviction.

    X:Artificial Analysis (@artificialanlys)First-partyAI score 40
Yesterday
  1. Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 on November 9. In a demo shared by Anthropic employee Jake Eaton on X, Fable 5.2 skipped image models entirely, writing Python code that lays down individual brush strokes on a virtual canvas, simulating brush texture, paint wetness and canvas grain and correcting mistakes with new strokes.

    IT之家AI score 53
Nov 15
  1. According to The Decoder, Anthropic's Claude Fable 5.2 does not produce images by computing pixels; instead it writes Python code that lays hundreds of thousands of simulated brushstrokes on a virtual canvas, complete with bristle texture and paint buildup. Anthropic employee Jake Eaton said a single photorealistic parking garage scene took roughly 700,000 strokes.

    The DecoderAI score 55
  2. Artificial Analysis testing shows that although Anthropic cut Claude Fable 5.2's per-token price by 20%, one task at max effort now costs $26.54 versus $12.99 on Fable 5.1, because the new model writes about 1.7 times as many output tokens. The post argues per-token price is the wrong metric and suggests testing your own workload at several effort levels before switching models on a pricing headline.

    DEV Community · AIAI score 59

Other stories people are talking about

How is heat calculated?About the method

Heat counts how many independent sources covered a story in the last 48 hours: one source counts once no matter how many posts it published, decaying with a 24-hour half-life. What ranks first is what many people are talking about.

This page aggregates public feeds. Headlines and summaries are machine-organized and remain the property of the original authors; verify important facts at the source. If you believe a headline or summary infringes your rights, email the contact address in the footer with the page URL and basis of your claim, and we will remove or replace it after review.

Surge
Discussion rising fast
New
First report within 6 hours
Rising
Still gathering discussion

Back to AI industry updates →