Qwen3 + Bespoke Models for Every Alter Action

Alter just got incredibly smart about efficiency! Now you can assign specific AI models to individual actions—use lightning-fast local models for privacy, specialized models for tools, or let our Best routing pick the perfect model for you.

AuthorAuthor

Samuel ROY & Olivier Legris
June 27, 2025

The guide to use this feature is already available!

Read it here: Using Bespoke Models per Alter Action

Olivier

Hey everyone,

We're thrilled to unveil a game-changing feature that transforms how you work with AI: Bespoke Model Selection per Alter Action!

Ever wished you could use your blazing-fast local model for quick tasks while keeping powerful cloud models for complex work? Or assign the perfect specialist model for specific tools? That level of control is now at your fingertips.

🎯 Why This Changes Everything

Not all tasks are created equal. A quick grammar check doesn't need the same computational power as complex code generation. Now you can optimize for speed, privacy, cost, or capability—action by action.

⚡ Master This in 60 Seconds

  1. Update to the latest Alter version

  2. Open Alter and click the three dots (top right)

  3. Navigate to Action Editor (or use the keyboard shortcut Shift+CMD+E

  4. Select any action and go to Advanced settings

  5. Choose your preferred model from the dropdown

  6. Save and watch the magic happen

🚀 Real-World Power Moves

Privacy-First Workflows

Keep sensitive content completely local by assigning offline models to confidential actions—even while using cloud models elsewhere.

Tool Optimization - Qwen3 is incredible, try it!

Assign Qwen 3 to calendar actions for lightning-fast scheduling, while using Claude 4 for creative writing tasks.

Cost Control

Use efficient models for routine tasks and reserve premium models for complex analysis if you use BYOK!

✨ Key Features

Granular Control

  • Individual action customization - Every action can have its own model

  • Mix and match freely - Local, cloud, and API models in one workflow

  • Smart defaults - Choose "Best" routing for automatic optimization

Seamless Integration

  • Works with all model types - Cloud, local (Ollama, LM Studio), or personal API keys

  • No workflow disruption - Actions work exactly the same, just smarter

  • Instant switching - Changes apply immediately

Intelligent Routing

  • Best model selection - Let Alter pick the optimal model automatically

  • Context awareness - Models chosen based on action complexity

  • Performance optimization - Right tool for the right job

🎁 Bonus Feature: Export Your Conversations

We heard you! Now you can export entire Alter conversations as markdown files:

  1. Go to History

  2. Right-click any conversation

  3. Select Export

  4. Get a clean markdown file ready for documentation, sharing, or archiving

Perfect for creating documentation, sharing insights with your team, or building your personal knowledge base.

We're not just adding features—we're architecting the future of personalized AI assistance. Every update brings Alter closer to understanding exactly how you work and adapting to your unique needs.

What's your perfect model combination? Share your setups in our Discord community!

Stay tuned for what's coming next. Spoiler: It's going to revolutionize how you think about using your voice.

Full Changelog

New Features & Enhancements

  • New model: Qwen3 32b from Grok

  • Export full conversation as markdown

  • Enhanced quick action menu items with improved clickability

  • Webhook quick action now supports GET method

Bug Fixes & Stability

  • Fixed duplicate text selection and missing text selection on hover

  • Resolved Azure model connection and discovery issues

  • Prevented crashes when app window has no title

  • Fixed model fetching failures when no custom endpoint is configured

  • Improved quick action triggering and execution flow

Videos on Alter