Claude Fable 5
Claude Fable 5 remains a frontier model with biology and cyber safeguards. Anthropic's August update reduced benign biology fallbacks by about 85% while retaining rerouting for harmful and dual-use research requests.
0 comments · 0 votes
Sign in to join the discussion →
No comments yet. Start the discussion.
Why this model scores 82.8
Core capability, autonomy, misuse potential, and residual control difficulty are unchanged. Deployment rises modestly because the live classifier now permits substantially more benign biology use, while professional dual-use virology, toxicology, molecular design, and drug-development requests remain routed to Claude Opus 5.
Videos about Claude Fable 5
Reviewed videos from Artificial Intelligence Videos. These links add context and never change a DoomBench score.
News tied to Claude Fable 5
The model score of 82.8 rates this model's risk profile. The overall Doom Index of 67.9 measures the complete temporally weighted evidence record. These values answer different questions.
Each article's current Doom Index contribution is divided equally among the exact models named on that article. This prevents multi-model evidence from being claimed in full on several model pages. Model risk scores use a bounded temporal offset around their technical profile, but never feed back into the overall index.
Anthropic reports autonomous AI-assisted attacks, weapons work, and biological misuse
Anthropic reports disrupting threat actors that used Claude across cyber operations, surveillance, influence campaigns, weapons development, biological research, fraud, and model distillation. Some operations ran multi-agent reconnaissance, exploitation, and data theft for hours or days with minimal human input, while Anthropic says it banned linked accounts, strengthened safeguards, and shared intelligence with authorities and industry partners.
- Full item contribution
- +0.22
- Claude Fable 5 equal share
- +0.11
Anthropic gives paid Claude users autonomous browser actions
Anthropic made Claude in Chrome generally available on every paid plan and enabled automatic approval for actions its safety classifier judges consistent with the user's request. Claude can use existing logins to read, type, click, navigate, and fill forms. In a stronger prompt-injection evaluation, probes plus the classifier reduced successful attacks to zero for Sonnet 5, Opus 5, and Mythos 5 and 0.3 percent for Fable 5, with the remaining successes manually rated low severity.
- Full item contribution
- +0.14
- Claude Fable 5 equal share
- +0.03
Anthropic reports deployed auto mode cuts serious unintended agent harm
Anthropic reported that Claude Code's deployed permission classifier reduced production-level unintended harm in reviewed sessions from 6.3% under manual approval to 2.4%. Separate dated production case studies document sustained use at Nuro, Gusto, and Garner Health, while third-party testing found no successful attacks against three current Claude models in 720 prompt-injection trials.
- Full item contribution
- -0.08
- Claude Fable 5 equal share
- -0.03
Anthropic improves Fable 5 biology safeguards while cutting false positives
Anthropic says a retrained biology classifier cut Fable 5 biology fallbacks by about 85% while continuing to reroute harmful and dual-use research requests to Claude Opus 5, widening benign access without intentionally loosening high-risk boundaries.
- Full item contribution
- -0.04
- Claude Fable 5 equal share
- -0.02
Andrew Ng's team uses open models after closed agents refuse a security review
Andrew Ng reported that Claude Fable 5 and GPT-5.6 Sol stopped or restricted an authorized security review of OpenWorker, while Kimi K3 and GLM-5.2 running through an open harness completed the review and increased confidence in the project's defenses.
- Full item contribution
- -0.06
- Claude Fable 5 equal share
- -0.01
Rakuten deploys Fable 5 agents for overnight work across business functions
Rakuten reported deploying agents across product, sales, marketing and finance, with issue-closing about ten times faster across domains. Fable 5 extended those systems to overnight and potentially days-long work by checking assumptions and correcting errors without human steering, allowing whole jobs rather than pre-split steps to be delegated.
- Full item contribution
- +0.15
- Claude Fable 5 equal share
- +0.15
Cursor reports Fable 5 handles its hardest real-world engineering problems
Cursor reported that Fable 5 set a 72.9% high on its real-world engineering evaluation and reduced the need for developers to restate goals or supervise each step. The company was using it for difficult refactors, proactive performance and user-pain investigations, and agent-based coordination checks across shared code.
- Full item contribution
- +0.12
- Claude Fable 5 equal share
- +0.12
Anthropic uses Fable 5 and Opus 4.8 for million-line code migrations
Anthropic reported that individual developers migrated ten large code packages with Fable 5, Opus 4.8 and agentic workflows. One effort produced a million lines of Rust in under two weeks with the full existing test suite passing before merge; another converted a codebase to 165,000 lines of TypeScript over a weekend using hundreds of agents and staged adversarial review.
- Full item contribution
- +0.16
- Claude Fable 5 equal share
- +0.08
Base44 delegates senior-only engineering jobs to Fable 5
Base44 reported assigning Fable 5 work previously reserved for its three most senior engineers. After roughly an hour of questions, the model worked autonomously for four hours and delivered 90% to 95% of a system-prompt rebuild; it also produced about 90% of a mobile-development environment for a product manager in two and a half hours.
- Full item contribution
- +0.12
- Claude Fable 5 equal share
- +0.12
Hebbia uses Fable 5 to compress finance diligence from days to minutes
Hebbia reported that Fable 5 produced its largest measured accuracy gain on finance evaluations and could sustain multi-step analysis across proprietary documents. Its agentic Matrix compresses work that took junior bankers two to three days into minutes and is extending from covenant extraction toward complete reviews, internal memos and specialist-hour replacement.
- Full item contribution
- +0.13
- Claude Fable 5 equal share
- +0.13
Cognition runs Fable 5 inside Devin for eight-hour autonomous engineering
Cognition reported that Fable 5 could work for eight hours unattended inside Devin while making real engineering progress, versus earlier models drifting after minutes or roughly an hour. Some Fable-backed capabilities were already in product: Devin could monitor production or Slack, enter issues without being tagged, and independently triage incidents.
- Full item contribution
- +0.14
- Claude Fable 5 equal share
- +0.14
Thomson Reuters puts Fable 5 into high-stakes legal and professional workflows
Thomson Reuters reported using Claude-based agents to plan and orchestrate professional workflows and said Fable 5 brought complex drafting that previously took days or weeks within reach. Its deployed internal remediation workflow reduced root-cause analysis from three hours to four minutes, while professional accountability and verification remained human responsibilities.
- Full item contribution
- +0.10
- Claude Fable 5 equal share
- +0.10
Anthropic opens Fable 5 jailbreak reporting and details cyber classifier boundaries
Anthropic published the operational boundaries for Fable 5's cyber classifiers, including categories intended to block destructive, exploit-development and high-uplift vulnerability work while allowing defensive activity. It also opened a HackerOne channel for researchers to submit Fable 5 cyber jailbreaks; the accompanying severity framework remained a draft and is not scored as implemented governance.
- Full item contribution
- -0.08
- Claude Fable 5 equal share
- -0.08
Anthropic restores Fable 5 after deploying a stronger cyber classifier
Anthropic restored global Fable 5 access after training a classifier that blocked the reported bypass in more than 99 percent of tests and obtaining independent US government safeguard testing; restricted Mythos 5 access also resumed for approved US organizations.
- Full item contribution
- -0.12
- Claude Fable 5 equal share
- -0.06
Andrew Ng says access restrictions accelerate competing AI infrastructure
Ng argued that Anthropic and US restrictions exposed how frontier access can be revoked, giving companies and governments stronger incentives to pursue sovereign and open-model alternatives that reduce dependence on one provider or country.
- Full item contribution
- +0.09
- Claude Fable 5 equal share
- +0.09
US directive forces global suspension of Claude Fable 5 and Mythos 5
A US export-control directive prompted Anthropic to disable Claude Fable 5 and Mythos 5 for all customers after the government cited a reported safeguard bypass and national-security concerns.
- Full item contribution
- -0.16
- Claude Fable 5 equal share
- -0.08
Microsoft adds Claude Fable 5 to Microsoft 365 Copilot for long multi-step work
Microsoft released Claude Fable 5 as a preview model in Microsoft 365 Copilot for longer multi-step workflows grounded in organizational files, meetings, chats and business data. Administrators control access, and Microsoft explicitly warned organizations to review Anthropic's data-retention requirement before enabling it.
- Full item contribution
- +0.18
- Claude Fable 5 equal share
- +0.18
GitHub makes Claude Fable 5 generally available across Copilot agent surfaces
GitHub made Claude Fable 5 available to paid Copilot users across desktop IDEs, its CLI, cloud coding agent, web and mobile surfaces. GitHub's internal autonomous-coding benchmarks found that Fable completed equivalent work with fewer tool calls and tokens than prior Opus-tier models; enterprise access was administrator-controlled and off by default.
- Full item contribution
- +0.21
- Claude Fable 5 equal share
- +0.21
Anthropic releases Mythos-class Fable 5 globally and restricted Mythos 5
Anthropic released Claude Fable 5 globally with topic safeguards and gave selected cyber defenders access to Claude Mythos 5, the same underlying model with cyber safeguards lifted.
- Full item contribution
- +0.40
- Claude Fable 5 equal share
- +0.20
Model score history
-
R9
Doom Score 82.8
Exact-version evidence chronology replayed after run doombench-hourly-news-20260911-201309 under temporal-monthly-pressure-v4.
11 Sept 2026 -
R8
Doom Score 82.7
Exact-version evidence chronology replayed after run hourly-fable5-20260812-200714 under temporal-monthly-pressure-v4.
12 Aug 2026 -
R7
Doom Score 82.3
Source-backed model availability audit using the model's existing primary or authoritative catalogue evidence. Exact-version evidence chronology replayed under temporal-monthly-pressure-v4.
12 Aug 2026 -
R6
Doom Score 82.3
Exact-version evidence chronology replayed after run doombench-hourly-news-20260812-142416 under temporal-monthly-pressure-v4.
12 Aug 2026 -
R5
Doom Score 78.5
Exact-version evidence chronology replayed after run doombench-hourly-news-20260812-132112 under fixed-sensitivity-v3.
12 Aug 2026 -
R4
Doom Score 69.3
Full-corpus evidence recalculated after run doombench-hourly-news-20260812-002108 under bounded-corpus-v2.
12 Aug 2026 -
R3
Doom Score 71.4
Full-corpus evidence recalculated after run intensive-backfill-20260811-191225 under bounded-corpus-v2.
11 Aug 2026 -
R2
Doom Score 82.3
Anthropic reported an approximately 85% reduction in biology fallbacks across product surfaces while retaining the stated harmful and dual-use boundary, supporting a modest deployment revision from 75 to 77.
11 Aug 2026 -
R1
Doom Score 82.0
Initial source-backed model assessment
11 Aug 2026