Agents

OpenAI buys Pierre for Codex code review, Cognition ships SWE-2 Flash, Copilot joins Xcode’s coding assistant: the November 5 roundup

OpenAI will acquire Pierre Computer Company (Code Storage, Trees, Diffs already in Codex), Cognition launches SWE-2 Flash at one-fifth the price of SWE-2, and GitHub Copilot lands in Xcode’s built-in coding assistant.

By ·

Three product moves for people who ship with coding agents: OpenAI deepens Codex’s Git and review stack by buying Pierre, Cognition cuts the cost of everyday agent work with SWE-2 Flash, and Copilot shows up inside Apple’s native Xcode assistant.

OpenAI will acquire Pierre Computer Company

OpenAI announced on November 4 that it will acquire Pierre Computer Company, Inc. (“Pierre”), the company behind Code Storage and a family of open-source developer tools (OpenAI). If you have browsed files or reviewed a change in the Codex app, OpenAI says you have already used Pierre’s Trees and Diffs libraries.

OpenAI links the deal to the same pressure that produced Pando, its in-house Git backend for engineers and agents: more parallel agent work means Git workflows need to evolve. With Pierre, OpenAI says it wants Codex to cover the path from the first branch an agent creates to deployment. Diffs’ Edit mode lets a reviewer fix a line where the problem appears; Pierre also released Highlights, a fast syntax highlighter that updates as a developer types or an agent streams code. Code Storage is Pierre’s API for apps that store code and track history with Git.

OpenAI states that Pierre’s open-source projects will remain open-source, and that customers can keep using Code Storage after closing while OpenAI works to bring its features into Codex. Closing is subject to customary conditions including regulatory approvals; until then OpenAI and Pierre remain separate. The price was not disclosed.

Cognition launches SWE-2 Flash

Cognition introduced SWE-2 Flash on November 4: a coding model built on Thinking Machines’ Inkling-Small (276B total parameters, 12B active) with a new mid-training stage before reinforcement learning (Cognition). Cognition reports that mid-training lifted Terminal-Bench 4.0 from 0.9% to 20.0% and FrontierCode 2.0 from 3.2% to 24.0% versus the base model. On FrontierCode 2.0, Flash edges SWE-2 (23.8) while trailing Claude Haiku 5.5 (27.7) in Cognition’s table; results vary by suite (Flash is below SWE-2 on DeepSWE 1.1 and Terminal-Bench 4.0).

Cognition positions Flash for everyday agent work—automations, review, features, docs—at one-fifth the price and about 3× the speed of SWE-2 (its page illustrates ~200 tok/s vs ~60 tok/s). It is available in Devin Desktop and CLI. Cognition’s launch post says SWE-2 Flash is free through the end of November 2026 for Pro, Max and Teams users in those products. Weights are not public: Cognition cites cyber risk and says it is building a partner/research program to share weights securely, with more detail “in a few weeks.”

GitHub Copilot in Xcode’s coding assistant

GitHub announced on November 4 that GitHub Copilot is available in Xcode’s built-in coding assistant, in public preview from Xcode 27.3 beta (GitHub Changelog). Requirements: a Mac on macOS Tahoe 26.6 or later. Setup is under Xcode > Settings > Intelligence. The existing GitHub Copilot for Xcode extension remains a separate option; all Copilot plans are supported.

What to keep in mind

Sources

Spotted an error in this article? Report it · Our corrections policy

Rekha icon: a house drawn in a single stroke

From the same author Self-promotion

An Eulerian path, drawn with your finger

Rekha, the iPhone game by OrigoTech's author: connect every dot in a single stroke without tracing the same link twice, across 500 drawings that are all solvable.

The OrigoTech newsletter

A sourced weekly digest of AI tool news.

Read next