Changelog
What changed, when, in plain words. The proof that the roadmap moves.
July 26, 2026
A new Studio surface brings AI image and video generation into Alias, Google (Nano Banana, Imagen, Veo), OpenAI (GPT-Image), xAI (Grok Image), and FLUX. We also standardized on a Western-only model roster: DeepSeek and other China-origin models have been removed.
July 16, 2026
When Alias breaks a big task into subtasks, each one now executes on the model chosen for it, a research subtask on a research model, a code subtask on a code model, instead of a single default. Autonomous runs get genuinely multi-model.
July 16, 2026
Every answer's compute cost is now accounted in dollars behind the scenes, spend guards protect heavy days, and the security layer around web access and streaming answers got stricter.
July 15, 2026
Chat history is now synced to your account, start a session on one device, pick it up on another. Private to you, with offline fallback if your connection drops.
July 9, 2026
The composer got a full refinement pass: color-coded modes (Chat, Guided, Research, Autopilot, Autonomous), a working ring that glows in the active mode's color, session history in the sidebar, and starter suggestions on the empty screen.
July 8, 2026
Alias now routes across Claude, GPT, Gemini, Grok, and Perplexity, with bring-your-own-keys if you have accounts, or ours if you don't. Plus a redesigned workspace and a model picker that explains each choice.
June 26, 2026
The workspace is in production: real accounts, real models, parallel multi-model answers, and the governed agent hand-off: the foundation everything above is built on.