1.3.0

RouterChat 1.3.0

This release is mostly about making RouterChat nicer to use. Settings has been redesigned, Write mode now opens right where you left off, and code blocks finally look like code.

New Features

• Code blocks have syntax colors, a language label, and a copy button, and stand out more from the text around them
• Attach any text or code file, not just the common types
• A search bar in the settings to quickly find what you need to change
• Each model in Settings > Models now shows its cached read price
• Dictation models now show their price in Settings > Dictation

Redesigns

• Settings is now redesigned to feel and look better
• The Write Mode landing page lets you pick up where you left off by showing your latest story, and a list of recent stories you worked on

Fixes/Changes

• Fixed Claude Haiku 5.5 failing with a "temperature is deprecated" error on a brand new install
• Fixed Claude Haiku 5.5 showing no price in the model list and no cost in Usage
• Fixed Claude Sonnet 5.5 cached reads costing double what they really do in Usage
• Fixed chat names and new lorebook entries failing with thinking models on OpenRouter, like GLM 5.3 Flash, before the model list had loaded
• Model search puts exact name matches first

Full Changelog: 1.2.3...1.3.0

1.2.3

RouterChat 1.2.3

This release fixes Claude models failing through OpenRouter for structured output features, and makes Write mode's lorebook updates safer and better. The lorebook now automatically retries edits that didn't go through. This can be disabled in Settings > Lorebook in write mode.

This is a beta release. It's been tested and is stable, but please report any bugs or issues you encounter on GitHub.

New Features

• The prompt bar settings button now displays the model provider next to the model name so you dont have to go all the way to settings to see which provider you are using

Fixes

• Claude models through OpenRouter no longer fail with "Provider returned error" on lorebook updates, chapter edits, entry generation, brainstorm, and the repair tools
• OpenRouter error messages now include the provider's own explanation so failures are easier to understand (in your RouterChat logs)
• Hidden entries are shown to the model so it stops recreating them, and it can bring one back into context when the story calls for it
• Only new or changed paragraphs are sent to the model, so the Timeline stops collecting duplicate events after small edits
• Running an update on a chapter with no changes since the last run makes no model call
• Skipped edits get one short retry with a fresh view of the lorebook, controlled by a new "Retry skipped edits" switch in story settings (on by default)

Full Changelog: 1.2.2...1.2.3

1.2.2

RouterChat 1.2.2

RouterChat now works with Anthropic directly. Add an Anthropic API key in Settings and you can use Claude models alongside OpenRouter, and switch between the two whenever you like.

This is a beta release. It's been tested and is stable, but please report any bugs or issues you encounter on GitHub.

New

• Anthropic provider: use Claude models with your own Anthropic API key
• Chats and stories keep using the provider they were started with, and the provider is shown on each one
• Stories can be moved to another provider with a move button found in the model picker
• Thinking and reasoning effort work with Claude models
• Provider tabs in Usage settings, with costs saved to a separate usage database (old costs are migrated automatically)

Fixes/Changes

• The composer now refuses images and files that are too big for the provider before you send them
• Updated the Terms of Service for Anthropic

Full Changelog: 1.2.1...1.2.2

1.2.1

RouterChat 1.2.1

RouterChat is back! This release brings a big round of redesigns across the app, prompt caching to help cut costs, a lot of brainstorm fixes, and RouterChat can now let you know when a new version is out. Support for more API providers is the next big goal, planned for 2.0.

(Same as 1.2.0, just fixed one critical bug in lorebook)

New

• Update checks: a dot appears next to the version number in the sidebar when a new release is out. It's on by default and can be turned off in Settings > Updates. RouterChat never sends usage data, and nothing is downloaded or installed automatically
• Prompt caching for chats and write mode, with cached reads shown in the Usage panel
• Setting to keep the prompt cache for 1 hour (on by default) and a toggle to turn prompt caching off
• Setting to hide batch models from the model picker
• The chat system prompt is now global for all chats and autosaves (uncaught bug was fixed)

Redesigns

• Settings is now a full page instead of a popup
• Lorebook is now a list with a side panel editor
• Brainstorm mode is easier to use and more visually appealing
• Write history got yet another makeover
• Terms of Service screen updated to be easier to read

Fixes/Changes

• Default model is now Sonnet 5
• The context wheel warns at 50, 75 and 90 percent full
• The Inter font is now bundled locally
• Fixed brainstorm showing empty after switching chapters
• Stopped or failed brainstorm prompts now show as cancelled or failed
• Navigation is blocked while a brainstorm is streaming
• Brainstorm rounds no longer overlap
• Fixed the saved default model not being used on page load
• Fixed the model name getting cut off in the phone composer
• Long lorebook entry titles now wrap instead of cutting off
• The "terms unavailable" popup now shows the real cause
• Updated the chat and write mode tours
• Fixed the lorebook entry box jumping the page while typing in long entries

Full Changelog: 1.1.7...1.2.0

1.2.0

RouterChat 1.2.0

RouterChat is back! This release brings a big round of redesigns across the app, prompt caching to help cut costs, a lot of brainstorm fixes, and RouterChat can now let you know when a new version is out. Support for more API providers is the next big goal, planned for 2.0.

This release updates the Terms of Service to cover update checks. You'll be asked to accept the revised terms after updating.

New

• Update checks: a dot appears next to the version number in the sidebar when a new release is out. It's on by default and can be turned off in Settings > Updates. RouterChat never sends usage data, and nothing is downloaded or installed automatically
• Prompt caching for chats and write mode, with cached reads shown in the Usage panel
• Setting to keep the prompt cache for 1 hour (on by default) and a toggle to turn prompt caching off
• Setting to hide batch models from the model picker
• The chat system prompt is now global for all chats and autosaves (uncaught bug was fixed)

Redesigns

• Settings is now a full page instead of a popup
• Lorebook is now a list with a side panel editor
• Brainstorm mode is easier to use and more visually appealing
• Write history got yet another makeover
• Terms of Service screen updated to be easier to read

Fixes/Changes

• Default model is now Sonnet 5
• The context wheel warns at 50, 75 and 90 percent full
• The Inter font is now bundled locally
• Fixed brainstorm showing empty after switching chapters
• Stopped or failed brainstorm prompts now show as cancelled or failed
• Navigation is blocked while a brainstorm is streaming
• Brainstorm rounds no longer overlap
• Fixed the saved default model not being used on page load
• Fixed the model name getting cut off in the phone composer
• Long lorebook entry titles now wrap instead of cutting off
• The "terms unavailable" popup now shows the real cause
• Updated the chat and write mode tours

Full Changelog: 1.1.7...1.2.0

1.1.7

RouterChat 1.1.7

RouterChat is back in active development! The top priority right now is adding support for more API providers.

Fixes/Changes

• Redesigned the sidebar
• New Chapter button now sits at the bottom of list of chapters in write mode
• Redesigned the Usage panel
• Fixed a slight flicker when closing settings modal

Full Changelog: 1.1.6...1.1.7

1.1.6

RouterChat 1.1.6

This marks the end of active development for RouterChat for the time being. The project is now in maintenance mode, with future updates limited to critical bug fixes, security fixes, dependency compatibility, and release infrastructure issues.
This release updates the documentation, support and security policies, and Terms of Service. You’ll be asked to accept the revised terms after updating.
Thanks for the ride, and thanks for using RouterChat.

I am currently planning to resume development in a few months.

Full Changelog: 1.1.5...1.1.6

1.1.5

RouterChat 1.1.5

New Features

• Added transcription
• Added transcription to the cost panel

Fixes/Changes

• Redesigned the Chat landing page
• Updated prompt bars across Chat, Write, and Brainstorm
• Improved how attached files are displayed

Full Changelog: 1.1.4...1.1.5

1.1.4

RouterChat 1.1.4

New Features

• Added a Usage panel in settings with spending, requests, and token usage for the last 7 days and lifetime totals per model

Fixes/Changes

• Improved handling of interrupted Write generations and disconnects to prevent lost model stream and edge cases
• Improved file upload and download safety
• Hardened website icon fetching against unsafe addresses and redirects
• Cleaned up frontend code to make future changes easier (no effect on user experience, just makes development easier)

Full Changelog: 1.1.3...1.1.4

1.1.3

RouterChat 1.1.3

New Features

• Added copy buttons for Brainstorm prompts and ideas
• Added a regenerate button for Brainstorm prompts

Fixes/Changes

• Redesigned Brainstorm for Write mode
• Fixed a bug so Brainstorm now properly keeps existing nodes until regeneration succeeds, instead of losing them if something fails
• Cleaned up the Write mode canvas
• Improved sidebar
• Changed prompt bar hover effects from changing brightness to using a pill style outline

Full Changelog: 1.1.2...1.1.3