# GPT-6 Astra

> GPT-6 Astra is OpenAI's newest flagship, released in September 2026 and sitting a full generation above the GPT-5.6 line. It is the model you reach for when the task is hard enough that a cheaper model burns more tokens failing than Astra costs succeeding — deep multi-step reasoning, large refactors, and agent loops that have to stay coherent over long horizons.

Canonical: https://routemux.com/openai/gpt-6-astra
Provider: OpenAI

## Key facts

| Field | Value |
| --- | --- |
| Model ID | openai/gpt-6-astra |
| Official model ID | gpt-6-astra |
| Context window | 1.1M |
| Max output | 128K |
| Input modalities | text, image |
| Output modalities | text |
| Capabilities | Vision, Reasoning, Tool Calling, Coding, Cache, Paid Cache Write, Strip Responses Summary Input |
| Upstream release | 2026-09-03 |
| Availability | available |

## Pricing

- RouteMux input: $1 per 1M tokens
- RouteMux output: $5 per 1M tokens
- Official input: $10 per 1M tokens
- Official output: $50 per 1M tokens
- Cache read: $0.1 per 1M tokens
- Cache write: $1.25 per 1M tokens
- Saving vs official: 90%
- Official price source: openai_x0.10, verified 2026-09-05

Billing is prepaid wallet, charged per successful request only.

## API access

Base URL: https://api.routemux.com

Supported protocols:

- openai_chat
- openai_responses
- anthropic_messages

Aliases that also route here: gpt-6, openai/gpt-6

## What is GPT-6 Astra

GPT-6 Astra is OpenAI's newest flagship, released in September 2026 and sitting a full generation above the GPT-5.6 line. It is the model you reach for when the task is hard enough that a cheaper model burns more tokens failing than Astra costs succeeding — deep multi-step reasoning, large refactors, and agent loops that have to stay coherent over long horizons.

It carries a 1.05M-token context window with 128K of output headroom, accepts text and image input, and exposes reasoning effort as a request parameter from low all the way up to max. Explicit prompt caching is supported, and cache reads are billed at a tenth of the input rate.

Through RouteMux it is $1 per 1M input tokens against OpenAI's $10, and $5 against $50 on output — a launch rate at a tenth of list price. The same model is reachable over three protocols: OpenAI Chat Completions, OpenAI Responses, and Anthropic Messages, so an Anthropic-shaped client can call it without rewriting its request layer.

## Key capabilities

- **1.05M-token context** — Large enough to hold a repository or a long document set in one request, with 128K of output headroom.
- **Reasoning effort up to max** — Effort is a request parameter spanning low, medium, high, xhigh and max — the same deployment can run cheap on easy turns and deep on hard ones.
- **Explicit prompt caching** — Cache breakpoints are supported, and cache reads bill at a tenth of the input rate.
- **Three protocols** — OpenAI Chat, OpenAI Responses, and Anthropic Messages all route to this model.

## Sources

- [OpenAI — Models](https://platform.openai.com/docs/models)
- [OpenAI — API pricing](https://developers.openai.com/api/docs/pricing)

_Content reviewed by Jerry Fan on 2026-09-05._