Cognition's Devin Integrates Claude Fable 5, Leads Real-World Engineering Benchmark

Cognition

Jun 10, 2026

Cognition has made Anthropic's Claude Fable 5 model available within its Devin AI software engineer across Cloud, Desktop, and CLI. This integration positions Fable 5 as the top performer on Cognition's FrontierCode benchmark, highlighting its advanced capabilities for production-grade code quality and mergeability in autonomous engineering tasks.

Cognition has integrated Anthropic's Claude Fable 5 model into its Devin AI software engineer, available across Devin Cloud, Desktop, and CLI. This model excels at long-horizon reasoning, debugging, and generalizing to unfamiliar tools, including MCP integrations and Computer Use.

FrontierCode Main Score (Claude Fable 5): 46.3%
Devin Ultra Agent Cost: ~40% more than default
Fable 5 Availability: Devin Cloud, Desktop, CLI
FrontierCode Evaluation Criteria: Quality, Mergeability
Fable 5 Capabilities: Long-horizon reasoning, debugging, unfamiliar tool generalization (MCP, computer use)

Following Anthropic's launch of Claude Fable 5 and its adoption by other AI coding tools like Cursor, the model now holds the #1 spot on Cognition's FrontierCode benchmark. This benchmark evaluates AI models on real-world engineering tasks, grading code quality and mergeability, confirming its production readiness.

Devin Cloud's Ultra agent now offers Claude Fable 5, tuned to cost approximately 40% more than the default Devin agent. The model is also accessible in Devin Desktop and CLI, enabling autonomous software development with advanced reasoning and debugging for complex tasks.

FrontierCode MAIN — Score; 4.8 Gemini 3.1 Flash Lite; 5.3 MiniMax M2.5; 5.5 SWE-1.6; 6.0 MiniMax M2.7; 6.9 Kimi K2.5; 15.1 Claude Sonnet 4.6; 16.0 Kimi K2.6; 16.7 Gemini 3.1 Pro; 17.8 GPT-5.4-mini; 23.0 Claude Opus 4.7; 25.5 GPT-5.5; 34.3 Claude Opus 4.8; 46.3 Claude Fable 5; Main is the 100 hardest FrontierCode tasks, dropping only the easiest tranche. Score (gated) is each trial's weighted rubric score only once it clears every blocker criterion, else 0 — so a run earns partial credit only after it is correct. Averaged over the selected tasks. — FrontierCode benchmark scores comparing various AI models, with Claude Fable 5 leading at a 46.3 percent score.

View the full update on devin.ai

Cognition

@cognition11h ago

Claude Fable 5 is now available in Devin. Fable 5 earns the #1 spot on FrontierCode, our benchmark for real-world engineering tasks that grades mergeability and quality:

1001.3k

View on X

Every HeadsUpAI update is written based on its original source and reviewed before it's published. Read our editorial standards →

Keep reading

Cognition Integrates GPT-5.5 to Power Longer Autonomous Engineering Sessions in Devin

Cognition added OpenAI's GPT-5.5 to its Devin autonomous software engineer as an Agent Preview. The update enables the agent to handle longer multi-step trajectories and resolve complex production bugs that previous models could not identify.

OpenRouter Adds Anthropic's Claude Fable 5 for Advanced Agentic Coding

OpenRouter3h ago

OpenRouter Adds Anthropic's Claude Fable 5 for Advanced Agentic Coding

OpenRouter has made Anthropic's Claude Fable 5 model available on its platform. This model is designed for complex, long-running coding and autonomous knowledge work, achieving state-of-the-art performance on various benchmarks. Its availability expands access to a frontier AI model for developers building agentic applications.

Cursor Adds Claude Fable 5, Sets New Coding Benchmark Record

Cursor3h ago

Cursor Adds Claude Fable 5, Sets New Coding Benchmark Record

Cursor, an AI-first code editor, has made Anthropic's Claude Fable 5 model available within its platform. The model achieved a new state of the art on CursorBench 3.1 with a score of 72.9%, surpassing the previous best by 8 points. This update signifies a notable improvement in AI coding capabilities for complex development tasks.

GitHub3h ago

GitHub Copilot Integrates Claude Fable 5 for Autonomous Coding

GitHub Copilot now offers Anthropic's Claude Fable 5, the first model in its Mythos class, for long-horizon, autonomous coding and knowledge-work tasks. This integration provides Copilot users with advanced agentic capabilities, enabling more efficient multi-step code resolution.