Reviews · NOVEMBER 9, 2026
Claude Fable 5.2 launches at $8/$40, with faster output and stronger coding scores
Anthropic’s new model offers a candidate for testing landing-page and lead-form changes—not evidence of higher customer conversion.
Anthropic released Claude Fable 5.2 and Claude Mythos 5.2 November 9, pricing Fable at $8 per million input tokens and $40 per million output tokens. TechCrunch reports input pricing fell 20% from $10.
Anthropic claims up to 30% faster output than their 5.1 predecessors. Optional fast mode costs $16/$80 per million input/output tokens for up to 2.5 times faster generation. That isn't measured development-time savings.
Coding scores depend on effort
Anthropic’s self-reported FrontierCode 2.0 results put Fable 5.2 at 49.1% with Xhigh effort, against Fable 5.1’s 32.4%, Opus 5.5’s 45.5%, GPT-6 Astra’s 42.8%, and Gemini 4 Carbon’s 41.1% at Medium effort.
At High effort, Fable scores 45.3%, approximately matching Opus’s best 45.5% at roughly half the cost per task. FrontierSWE v2 scores are 69.9%, versus 56.3% for Fable 5.1 and 62.3% for Opus 5.5.
The table uses highest available effort unless noted, with Fable and Opus safeguards enabled unless specified otherwise. Medium is default in Claude apps and Claude Code. FrontierCode measures code-change acceptance by open-source maintainers, not inquiries. These aren't independent website-productivity measurements.
Anthropic still recommends starting most coding and knowledge tasks with Opus 5.5, adjusting for complexity. See earlier Opus 5.5 and GPT-6 release coverage.
Internal results aren't conversion tests
Anthropic says Claude writes over 90% of merged core-codebase changes with human guidance. Agents using an earlier Fable 5.2 migrated a production service from Python to Rust; startup is now about 26 times faster. That's an internal deployment, not a landing-page test.
In Anthropic’s release, Joe Haddad, Member of Technical Staff at Vercel, said:
"Projects we’ve been pushing along with other models for weeks got finished and approved by our teammates in a few days with Claude Fable 5.2."
The testimonial reports neither spending nor a controlled comparison.
AWS access comes with review requirements
Amazon Web Services (AWS) confirms launch-day availability on Amazon Bedrock and Claude Platform on AWS. Fable is generally available; Mythos requires admission through Anthropic’s verification programs. Anthropic describes the same underlying model, with Fable’s cybersecurity and biology safeguards permitting broader access.
AWS supports the Anthropic Messages API, Bedrock Invoke API and Bedrock Converse API. Its example uses global.anthropic.claude-fable-5-2; both models support Bedrock US Geo CRIS and Global CRIS inference profiles. Claude Code’s /claude-api migrate updates model references, thinking settings and deprecated parameters.
AWS says Fable requires retention up to 30 days and human safety review by default. Amazon personnel review within the AWS boundary on Bedrock; users must set aws_review before invocation. Enterprise Frontier Safeguards supports zero data retention; AWS says eligible customers can use Fable 5.2 that way through June 30, 2027.
For small-team owners, a defined landing-page or lead-form change suggests a test, not proven savings: use synthetic submissions and check retention before real prospect data. Conversion improvement needs separate measurement on the business’s site.