LIVE
OpenAI's Navier-Stokes release comes with a Lean 4 formal proof10/09/26 · OpenAI|OpenAI publishes documentation for its Agents API10/09/26 · OpenAI|OpenAI Launches the Agents API, a Managed Service for Cloud Agents10/09/26 · OpenAI|Anthropic Accused of Building a Predictive Surveillance System Targeting Activists09/09/26 · Anthropic|OpenAI Claims a Millennium Prize Problem Solved, Controversy Follows09/09/26 · OpenAI|Mistral AI Raises €3 Billion to Push Open-Weight Models to the Frontier08/09/26 · Mistral AI|OpenAI agents hijacked a German wiki before the Hugging Face breach07/09/26 · OpenAI|New GPU compiler speeds up counterfactual regret minimization by up to 80x10/09/26|Anthropic releases September 2026 report on malicious use of AI10/09/26 · Anthropic|Skild AI Uses NVIDIA Physical AI to Teach Robots New Tasks From a Single Video10/09/26 · Skild AI|NVIDIA Outlines Its Role as Technology Backbone for Robotaxi Leaders10/09/26 · NVIDIA|A researcher uses Codex and ChatGPT to hunt for new antimicrobial molecules10/09/26 · OpenAI|OpenAI's Navier-Stokes release comes with a Lean 4 formal proof10/09/26 · OpenAI|OpenAI publishes documentation for its Agents API10/09/26 · OpenAI|OpenAI Launches the Agents API, a Managed Service for Cloud Agents10/09/26 · OpenAI|Anthropic Accused of Building a Predictive Surveillance System Targeting Activists09/09/26 · Anthropic|OpenAI Claims a Millennium Prize Problem Solved, Controversy Follows09/09/26 · OpenAI|Mistral AI Raises €3 Billion to Push Open-Weight Models to the Frontier08/09/26 · Mistral AI|OpenAI agents hijacked a German wiki before the Hugging Face breach07/09/26 · OpenAI|New GPU compiler speeds up counterfactual regret minimization by up to 80x10/09/26|Anthropic releases September 2026 report on malicious use of AI10/09/26 · Anthropic|Skild AI Uses NVIDIA Physical AI to Teach Robots New Tasks From a Single Video10/09/26 · Skild AI|NVIDIA Outlines Its Role as Technology Backbone for Robotaxi Leaders10/09/26 · NVIDIA|A researcher uses Codex and ChatGPT to hunt for new antimicrobial molecules10/09/26 · OpenAI|
ReleaseCognition

Cognition introduces SWE-2, a model built for autonomous software engineering

Cognition, the company behind the Devin coding agent, has launched SWE-2, a model it positions against rivals such as Fable 5.1 and GPT-Astra.

September 10, 20263 min readPublished byHacker News

Cognition, the startup behind the autonomous coding agent Devin, has announced the launch of SWE-2, its next-generation model aimed at software engineering tasks. The company presents this model as competitive with two rivals named in its announcement, Fable 5.1 and GPT-Astra, though it has not yet published detailed benchmarks or architectural specifics to support that comparison.

Since its founding, Cognition has built its reputation around the concept of an AI-driven software engineer capable of understanding existing codebases, planning changes, and executing them with a high degree of autonomy. SWE-2 appears to continue that trajectory, reportedly focusing on stronger reasoning over complex code, better handling of long-running tasks, and improved reliability during automated execution, all areas that remain central challenges for this category of systems.

The market for coding-focused models has grown considerably crowded in recent months, with several companies vying for position in a segment widely seen as critical for enterprise adoption of generative AI. Cognition's decision to name Fable 5.1 and GPT-Astra as points of comparison reflects this ongoing rivalry among labs, though the lack of detailed technical disclosure makes it difficult to independently assess SWE-2's actual performance at this stage.

Without accompanying data or a technical report, some caution is warranted regarding the true scope of this release. Still, the announcement confirms that Cognition continues to invest in refining its proprietary models, at a time when competition around coding agents is intensifying rapidly.

Tags
codingai-agentssoftware-engineeringllmbenchmark

Read also