Wayland 0.11.3

Every key routes to its own endpoint

Every connected provider, from Sakana and OpenRouter to the whole OpenAI-compatible family, lands on the right endpoint via the v2 registry bridge at spawn.

Sakana AI joins as first-class

Sakana AI (Fugu) arrives as a first-class provider, with its Fugu models and brand logo right where you expect them.

Live model catalogs

ChatGPT subscription, MiniMax, and DashScope stream their model lists straight from the provider, so you always see the current lineup.

Native ChatGPT subscription chats

ChatGPT subscription chats run through the engine's native ChatGPT provider and are selectable directly from the model picker.

Keyless local Ollama

Your local Ollama models appear in the picker and run keyless, so on-device models are one click away.

Smarter model guidance

Models without tool support point you to a tool-capable one instead of stalling a turn, and Claude Code starts on Opus when no model is pinned.

Global Wayland Memory in chat

Global Wayland Memory is wired into chat, so the assistant recalls you and your projects across conversations.

Cleaner approvals and onboarding

Calmer approval prompts, contained file preview, smoother back-after-error navigation, sharper onboarding and Doctor, plus MCP OAuth sign-in with timeout and cancel.

Added

Sakana AI (Fugu) as a first-class provider. Sakana AI joins the lineup with its Fugu models and brand logo, ready to connect alongside everything else.
Live model catalogs. ChatGPT subscription, MiniMax, and DashScope stream their current model lists straight from each provider, so the picker always reflects the live lineup.
Native ChatGPT subscription chats. ChatGPT subscription chats run through the engine's native ChatGPT provider and are selectable directly from the model picker.
Keyless local Ollama. Your local Ollama models appear in the picker and run keyless for instant on-device inference.
Global Wayland Memory in chat. Global Wayland Memory is wired into chat, so the assistant recalls you and your projects across conversations.
Flux Voice configuration block. Flux Voice gets its own dedicated config block for clearer, more reliable speech-to-text setup.

Improved

Every key routes to its own endpoint. Every connected provider, from Sakana and OpenRouter to the whole OpenAI-compatible family, lands on the right endpoint via the v2 registry bridge at spawn.
Smarter model guidance. Models without tool support point you to a tool-capable one instead of stalling a turn, and Claude Code starts on Opus when no model is pinned.
Cleaner approvals and preview. Calmer approval prompts, contained file preview, and smoother back-after-error navigation make everyday flows feel lighter.
Sharper onboarding and Doctor. Onboarding and the built-in Doctor read more clearly, with Windows npm and WSL CLI detection added.
MCP OAuth sign-in with timeout and cancel. Connecting an MCP server over OAuth now offers a clean timeout and cancel, so sign-in stays in your control.
Bundled engine updated to v0.12.8. This release bundles the latest Wayland Core engine for steadier routing and context handling.

Under the hood: Wayland Core v0.12.8. This release bundles the v0.12.8 engine, and the latest Wayland Core adds the Sakana AI (Fugu) provider, isolated profiles, an active-window kernel that sizes context accurately after a model swap, smart Flux-aware auto-compaction with memory handoff, JSON-stream observability with active-window percentage and structured traces, an OpenAI image default of gpt-image-1, DeepSeek v4-flash with 1M context, and a round of Windows reliability and security hardening. Explore Wayland Core →

Eight reasons to update.