Anthropic's Claude Fable 5 variant has landed inside Cognition's Devin agent and taken the lead on the company's new FrontierCode benchmark, which grades models on whether their code actually merges cleanly into real projects instead of just passing isolated tests.
Super Dario#1794
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:
The integration lets Devin users tap Fable 5 for engineering tasks right now, though rollout details and exact performance numbers remain tied to Cognition's announcement.
FrontierCode's emphasis on mergeability and overall quality is still early, so it is unclear how widely teams will adopt it versus existing coding benchmarks.
Positive users celebrate Claude Fable 5's FrontierCode benchmark wins and Devin integration for showing rapid real progress, while negative users dismiss the results as biased or self-measured.
Took 1 day for AI to 2x the score on the hardest programming benchmark ever made
@NickADobos I hope that it gets solved in the next 6 months and then we can move on to even more challenging tasks!
You can try Claude Fable 5 as part of Devin Cloud’s Ultra agent. Devin Ultra is our smartest and most capable agent, which excels at long-horizon tasks and debugging.
We tuned the harness so Ultra costs only ~40% more than default Devin agent.
Claude Fable 5 is also available in Devin Desktop and Devin CLI.
A new top scorer just one day after our benchmark released! Especially strong on the hardest tasks: 13.4% -> 29.3% on FrontierCode Diamond compared to Opus 4.8.
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:
Mythos 5 / Fable 5 blows away the competition on @cognition FrontierCode Diamond (newest and least saturated SWE benchmark)
FrontierCode tasks 👇
Try out Devin today! http://devin.ai
Read our blog post: https://devin.ai/blog/claude-fable-5-available-in-devin
@cognition https://cognition.ai/blog/frontier-code
Mythos 5 / Fable 5 blows away the competition on @cognition FrontierCode Diamond (newest and least saturated SWE benchmark)
FrontierCode tasks 👇
@cognition Are we already gonna get saturated 😭
@cognition 🔥
@cognition Gaslighting benchmarks. Real world 5.5 mogs 4.8.
@cognition LETS GOO
@cognition @devarbol definitely not a bubble. Another step change in capability.
@cognition Finally, Devin's user will get a taste of Fable 5
@NickADobos Don’t underestimate the power of a goblin
@cognition OpenAI needs to cook something big now to compete 💀
@cognition benchmark stacking gets boring but fable 5 actually sounds useful for once
is mergeability the metric that matters most here?
@cognition oh boy
@cognition Mirror, Mirror On the Wall, Who is the Mightiest Model of Them All...
@cognition Wow, that's a significant jump up. Excited to build with Fable 5
@cognition Waiting for people to realise this result here
Claude Fable 5 is fabulous
Anthropic's Claude Fable 5 variant has landed inside Cognition's Devin agent and taken the lead on the company's new FrontierCode benchmark, which grades models on whether their code actually merges cleanly into real projects instead of just passing isolated tests.
Super Dario#1794Claude 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: