Cline released CLI version 3.0.6 with updated ChatGPT provider model list. The patch adds support for codex variants and three new GPT-5 series models: gpt-5.2, gpt-5.4, and gpt-5.4-mini.
Google is rolling out a new "Thinking level" option in the Gemini app, allowing users to toggle between Standard and Extended modes when using Gemini 3 Flash or Gemini 3.1 Pro. The app is also preparing integrations with Canva, Instacart, and OpenTable, expanding beyond its current third-party connections.
Recent open-weight LLM releases from Google, DeepSeek, and others are adopting architectural techniques that reduce KV cache size by approximately 50% at long contexts. These include cross-layer KV sharing in Gemma 4, which saves 2.7 GB at 128K context for the E2B model, and compressed convolutional attention in ZAYA1-8B.
Microsoft Research has released Fara-7B, a 7-billion parameter small language model designed for computer automation tasks. The model, which took 2.5 days to train on 64 H100 GPUs, can navigate websites to complete tasks like booking restaurants and shopping, using screenshots as input with a 128K token context window.
Augment Code benchmarked its Auggie agent against Claude Code on Claude Opus 4.7, reporting a 67.4% pass rate versus 66.3% while cutting costs by 33%. The company attributes savings to a semantic context engine that reduces cache read tokens by 32% and output tokens by 37% compared to Claude Code's keyword-based retrieval.
Google's new Gemini Intelligence features require 12GB of RAM, Gemini Nano v3 or higher, flagship chips, and 5 years of OS upgrades. The requirements exclude recent devices including the entire Pixel 9 series and Samsung's Galaxy Z Fold 7, which run Gemini Nano v2.
Replit released its first iPhone app update in four months after resolving App Store review issues with Apple. The update brings Agent 4, the company's latest AI coding assistant, along with parallel agent support and cross-workspace project viewing.
OpenAI is rolling out a preview feature allowing ChatGPT Pro users in the US to connect their bank accounts through Plaid for personalized financial advice. The integration provides access to 12,000+ financial institutions and generates visual dashboards of user finances.
GitHub is piloting an experimental AI agent designed to automate accessibility testing and remediation. The tool aims to help developers identify and fix accessibility issues in their code and user interfaces without requiring specialized expertise.
GitHub is piloting an experimental accessibility agent that uses AI to help users with disabilities navigate software interfaces. The company disclosed the project in a technical blog post detailing lessons learned during development.
Anthropic has released Claude for Small Business, a collection of 31 AI skills for Claude Cowork subscribers. The standout feature is /review-contract, which analyzes legal contracts and flags problematic clauses in approximately five minutes. The tool requires at minimum a $20/month Claude Pro subscription.
xAI has released Grok Build, a coding agent and CLI tool positioned to compete with Anthropic's Claude Code and other AI coding assistants. The early beta is available exclusively to SuperGrok Heavy subscribers at $300 per month.
GitHub has expanded its Copilot usage metrics API to include team-level reporting. The new user-teams endpoint maps each Copilot-licensed user to their team memberships, allowing organizations to analyze AI coding assistant adoption and usage patterns across team structures.
Security researchers at Calif used Anthropic's Mythos Preview model to develop a working macOS kernel memory corruption exploit on M5 silicon in five days, bypassing Apple's Memory Integrity Enforcement (MIE) system. The exploit chain targets macOS 26.4.1 and escalates from unprivileged local user to root shell using two vulnerabilities and several techniques.
OpenAI has integrated its Codex coding agent into the ChatGPT mobile app for iOS and Android, allowing developers to monitor live development environments and manage workflows from their phones. The update, announced May 14, 2026, is now available in preview across all ChatGPT plans.
OpenAI has integrated remote Codex control into the ChatGPT mobile apps for iPhone and Android. Users can now approve tasks, review outputs, and manage Codex running on Mac computers, laptops, or remote environments directly from their smartphones.
OpenAI is rolling out mobile access to Codex, its desktop AI tool that can write code and control applications on users' computers. The feature allows users to control Codex on their desktop from the ChatGPT iOS and Android apps, with real-time updates flowing back to mobile devices.
Microsoft is removing Claude Code access from its Experiences + Devices division by June 30, 2026, redirecting thousands of engineers to GitHub Copilot CLI instead. The decision follows six months of Claude Code proving more popular than Microsoft's own coding tool among internal developers.
IBM released two new multilingual embedding models under Apache 2.0: a 97M-parameter compact model scoring 60.3 on MTEB Multilingual Retrieval (highest in its size class) and a 311M full-size model scoring 65.2. Both support 200+ languages with enhanced retrieval for 52 languages, handle 32K-token context (64x increase over predecessors), and include code retrieval across 9 programming languages.
A month after its initial release, a newer checkpoint of Anthropic's Mythos Preview became the first model to complete the UK AI Safety Institute's 'Cooling Tower' cyber range, solving it in 3 of 10 attempts. The model also completed 'The Last Ones' range in 6 of 10 attempts, surpassing OpenAI's GPT-5.5 and demonstrating capability improvements within a single model version.