GPT-6 Astra and GPT-6 Astra Pro API: now on FastMetal
OpenAI's GPT-6 Astra (gpt-6-astra) and GPT-6 Astra Pro (gpt-6-astra-pro) are available on FastMetal. Astra is OpenAI's flagship after the GPT-5.6 series, aimed at long-running tasks and deep analysis.
Capabilities
- 1.05M-token context: enough for a large codebase or a long document set in one request.
- Image and file input alongside text.
- Reasoning effort per request: set
reasoning.efforttolow,medium,highorxhigh. With nothing set, it usesmedium. Reasoning tokens are billed as output tokens, solowkeeps the bill down when a short answer is enough. See the reasoning guide. - Astra Pro is GPT-6 Astra run with
reasoning.modeset topro. It is the same model, tuned to put answer quality first on hard problems.
Calling it
The API is OpenAI-compatible:
curl https://api.fastmetal.ai/v1/chat/completions \
-H "Authorization: Bearer $FASTMETAL_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "gpt-6-astra",
"messages": [{"role": "user", "content": "Hello"}],
"reasoning": {"effort": "low"}
}'
For Astra Pro, change model to gpt-6-astra-pro. The endpoint, the API key and your balance stay the same. To try either without code, pick it in chat.
Pricing
Both are billed per million input and output tokens, from a prepaid balance. Current rates are on pricing and the model pages in the catalog.
How it compares
GPT-5.6 Sol vs GPT-6 Astra puts both models' real answers to the same prompts side by side. For a head-to-head against Anthropic's flagship of the time, see GPT-6 Astra vs Claude Fable 5.1.