From the source

Rebuilding Devin for Claude Sonnet 4.5: Lessons and Challenges

Capability change

Image published with the post by Cognition — Image from cognition.com

Cognition rebuilt Devin for Claude Sonnet 4.5, resulting in a 2x faster and 12% better performance on Junior Developer Evals.

The new version is available in Agent Preview, while the old version remains available.
The post details observations about the model's awareness of its context window, note-taking behavior, and parallel tool execution.

From the source
We rebuilt Devin for Claude Sonnet 4.5. The new version is 2x faster, 12% better on our Junior Developer Evals, and it's available now in Agent Preview.
cognition.com

Most covered

OpenAI OpenAI acquires Orchestra Bio
The Decoder: “OpenAI acquires Orchestra Bio to build AI agents that speed up drug development”
Reporting · 18 outlets
Image from openai.com
OpenAI· 20 Oct
OpenAI acquires Orchestra Bio
DeepSeek 🚀 Introducing DeepInfer, our in-house inference framework, now fully open source!
SCMP: “DeepSeek seeks ‘total victory’ over closed-source AI with new open-source tool”
Reporting · 14 outlets
DeepSeek· 20 Oct
🚀 Introducing DeepInfer, our in-house inference framework, now fully open source!
OpenAI ChatGPT Ads expands to 26 new countries as over 100,000 advertisers have used the platform
The Information: “OpenAI Expands ChatGPT Ads to 26 More Countries, Says It Has More Than 100,000 Advertisers”
Reporting · 12 outlets
Image from openai.com
OpenAI· 21 Oct
ChatGPT Ads expands to 26 new countries as over 100,000 advertisers have used the platform
Meta Inside Prometheus: Meta’s Journey to a Gigawatt-Scale Supercluster
RuntimeWire: “Meta says Prometheus delivered 85% goodput while training Muse Wonder”
Reporting · 7 outlets
Image from engineering.fb.com
Meta· 19 Oct
Inside Prometheus: Meta’s Journey to a Gigawatt-Scale Supercluster
Inception Diffusion at Scale: Introducing Mercury 3
The Decoder: “Inception's Mercury 3 diffusion model claims to beat GPT-6 Luna and Claude Haiku 5.5 at a fraction of the wait”
Reporting · 6 outlets
Image from inceptionlabs.ai
Inception· 19 Oct
Diffusion at Scale: Introducing Mercury 3

Other stories

Other stories

Other stories

Rebuilding Devin for Claude Sonnet 4.5: Lessons and Challenges

From the source

Other stories

Building in London

National Design Studio and Cognition Launch the American Forge Residency

Memory and dreaming: how Devin learns from working with you

Devin Mobile is now available in beta.

Cognition Crosses $1B in Annualized Revenue Run Rate

Building the Future of Software Engineering in Latin America