Google has published its July 2026 'Gemini Drop' roundup, covering voice control on macOS, a worldwide rollout of Gemini Spark, new Flash models, and expanded app integrations.
xAI added image and voice reference support to its "Grok Imagine Video 1.5" model, along with text-only video generation and native 1080p output.
GitHub will deprecate six models across all Copilot experiences on September 1, 2026, including Gemini 3.1 Pro, Claude Opus 4.5/4.6, Claude Sonnet 4.5/4.6, and Raptor Mini.
OpenAI has improved the price and speed of its GPT-5.6 lineup, cutting the lightweight Luna model by 80% and the balanced Terra model by 20%, while adding a 2.5x-faster "Fast mode" for the flagship Sol model.
GitHub has shipped the July 2026 update to GitHub Copilot in Visual Studio, adding a new agent (preview) built on the Copilot SDK, built-in skills authored by the .NET and Azure teams, and improved code review.
Cognition has launched Stacked PRs in Devin, automatically breaking large AI-generated changes into a stack of small, independently reviewable pull requests that rebase themselves as feedback comes in.
Perplexity has rebranded and upgraded its collaboration feature Spaces into Projects, adding a persistent file system where people and AI agents can edit shared files directly, plus a self-improving memory that learns from project history.
Alibaba's Tongyi Lab has released a new automatic speech recognition (ASR) lineup, "Qwen-Audio-3.0-ASR," as a hosted API. It ships streaming and file-transcription variants covering China's seven major dialect groups, 20+ regional accents, and 30 languages overall.
Runway's developer platform now offers ElevenLabs' Eleven v3 model through its text-to-speech endpoint, letting developers generate expressive speech with inline audio tags like [laughs] and [whispers], alongside the same sunset of Gen-3 Alpha Turbo and Gen-4 Aleph.