GPT-6 Astra

At a glance

GPT-6 Astra (API id gpt-6-astra) is a reasoning model with function calling and streaming and a 1,050,000-token context window. OpenAI prices it at $10 per million input tokens and $50 per million output tokens - its premium tier, above GPT-5.6 Sol.

A GPT-6 Astra reply costs 5 credits, the same tier as Claude Opus and GPT-5.6 Sol. If you use your own OpenAI key instead, OpenAI bills you at the API price.

What it is good at

Hard, multi-step work: deep research, long and complex documents, and goals that take many tool calls to finish. Magnus keeps its reasoning effort low by default so chat stays quick and reasoning tokens stay cheap; the model still thinks harder when a task demands it.

• For everyday chat and scheduled background tasks, GPT-5.6 Luna (1 credit) is the better default - switch to GPT-6 Astra for the work that earns its price.

Using it on Magnus

GPT-6 Astra is in the model picker for OpenAI keys and on the All-inclusive plan - nothing to enable. Switching models never resets your workspace: goals, memory, documents, and integrations carry over untouched.

1. Open Settings and pick OpenAI as your provider (it is included in your plan).
2. Select GPT-6 Astra in the model picker - existing agents keep their history when you switch models.
3. Give it the hard problems: research briefs, complex documents, and multi-step goals.