Platform Changelog
Updates to the Microbix Neural Gateway, Aether OS, and CLI tools.
Extended Thinking Models
Introduced extended reasoning capabilities allowing the models to think for multiple minutes to solve complex logical problems before returning the response. Perfect for advanced math, heavy coding tasks, and robust agentic pipelines.
Voice Synthesis & Input
Added low-latency native voice communication to Aether OS. Speak directly to your agent, and it will respond with expressive, human-like voice synthesis using our latest audio models.
Discord & Slack Bot Integration
Deploy Microbix agents natively inside your team's communication platforms with just a few clicks from the dashboard. Context-aware bots that read your team's history and answer questions automatically.
Neural Gateway 2.0
Complete rewrite of our core routing engine. Failovers are now 3x faster, with sub-50ms latency across global edge locations. Added automatic request retry queues and advanced load shedding.
Multi-Model Routing API
New endpoint allows developers to specify fallback models. If Ultra fails or rate-limits, it seamlessly routes down to Pro or Fast without throwing an error to your application.
CLI Approval Workflows
Aether Code CLI now supports manual approval steps for destructive actions. When the agent wants to delete a file or run a dangerous command, it stops and asks for 'y/n' confirmation.
Usage-Based Billing API
Track your mX credit consumption in real-time. Added `/v1/usage` endpoint to programmatically fetch your organization's daily token spend and credit balances.
Aether Ultra Flagship Model
Launch of our most capable model yet. Aether Ultra provides top-tier performance on complex reasoning tasks, code generation, and long-context understanding (up to 200k tokens).