Claude Fable 5.1 Reaches GitHub Copilot on Day One, Requiring Data Retention by Default
Anthropic's new coding-focused Claude Fable 5.1 became selectable in GitHub Copilot the day it launched, but the model requires data retention by default and stays off until administrators enable it.
Overview
Anthropic launched Claude Fable 5.1 alongside a more restricted sibling, Claude Mythos 5.1, describing the pair as “the world’s most advanced models for coding and knowledge work,” according to Anthropic. GitHub made Fable 5.1 selectable in GitHub Copilot’s model picker the same day, calling it the latest model in Anthropic’s Mythos class, designed for “long-horizon, autonomous coding and knowledge-work tasks,” according to GitHub Blog. The rollout comes with a catch for enterprise developers: the model requires data retention by default, and admins have to turn it on before their teams can use it.
What We Know
Fable 5.1 is available to Copilot Pro+, Max, Business, and Enterprise users, and can be selected from the model picker across Visual Studio Code, Visual Studio, Copilot CLI, GitHub Copilot coding agent, the GitHub Copilot app, github.com, GitHub Mobile on iOS and Android, JetBrains IDEs, Xcode, and Eclipse, according to GitHub Blog.
The model ships with an operational wrinkle that sets it apart from other Claude models on Copilot: it requires data retention by default. GitHub’s changelog states that when Fable 5.1 is used, “Anthropic retains data, including prompts and outputs, to operate safety classifiers that detect harmful use,” according to GitHub Blog. Eligible enterprises can still run the model with zero data retention through the end of the calendar year under what GitHub describes as a time-bound exemption while Anthropic rolls out a program called Enterprise Frontier Safeguards, according to GitHub Blog. On the Copilot Business and Enterprise plans, administrators must manually enable the Claude Fable 5.1 policy in Copilot settings, since the policy is off by default, according to GitHub Blog.
Anthropic frames Enterprise Frontier Safeguards as a way to give customers “complete privacy while still being state-of-the-art at preventing adversarial use,” according to Anthropic.
On pricing, Anthropic says Fable 5.1 will cost an estimated 25% less than Fable 5 for typical workloads, and that cache reads now cost 75% less, at $0.25 per million tokens, according to Anthropic. SiliconANGLE reported separately that an improved prompt-caching mechanism lets Fable 5.1 run typical workloads roughly 25% more cost-efficiently than its predecessor, with agentic applications potentially seeing savings of up to 45%, according to SiliconANGLE.
On coding performance, SiliconANGLE reported that Fable 5.1 completed Terminal-Bench-Science 0.1, a benchmark measuring how well models complete research tasks, with a score of 52.6% — more than double what the prior-generation Fable 5 model achieved, according to SiliconANGLE. The article also reported that Fable 5.1 scored higher on the Terminal-Bench 4.0 coding benchmark than its predecessor, without a precise figure.
Mythos 5.1, the sibling model announced alongside Fable 5.1, is not part of the Copilot rollout. Anthropic says it “is available only through our trusted access programs,” according to Anthropic. SiliconANGLE reported that Mythos 5.1 supports advanced tasks such as finding exploit paths and quoted Anthropic describing it as having “the strongest cyber capabilities of any model we’ve released,” according to SiliconANGLE.
What We Don’t Know
Neither GitHub nor Anthropic has published an exact date for when Enterprise Frontier Safeguards will fully replace the current default-data-retention arrangement, only that the zero-data-retention exemption runs through the end of the calendar year. The precise size of Fable 5.1’s improvement on the Terminal-Bench 4.0 coding benchmark was not clearly reported and is omitted here rather than guessed at.