🎉 Network upgrade complete — three API endpoints are now available

We have completed a network upgrade. Users in different countries and regions can now choose the more stable API endpoint:

• Main route: https://modelflare.dev/v1 — recommended for everyday use, with support for longer non-streaming requests
• Cloudflare Global Acceleration: https://cf.modelflare.dev/v1 — optimized for most regions
• Origin (EdgeOne): https://origin.modelflare.dev/v1 — accelerated for Asia and mainland China, and can be used without a VPN

Test latency on the console home page and choose the faster, more stable route for you.

🎉 GPT-6 Astra is now live — try the world's strongest model

OpenAI's latest flagship gpt-6-astra is now available on Modelflare. This is currently the world's most powerful model. You're welcome to try it.

Official price per million tokens: input $10, output $50, cache read $1.

🎉 GLM 5.3 Flash is now live, with lower GLM and Kimi group prices

Modelflare now supports glm-5.3-flash on glm-stable, with a 1M context window. It works with OpenAI Chat Completions, Responses, and Anthropic Messages.

After further optimization, we have also cut GLM and Kimi group prices:

1. kimi-stable is now 0.65x
2. glm-stable is now 0.5x

For glm-5.3-flash, current effective pricing is:

• Input: $0.075 / 1M tokens
• Output: $0.25 / 1M tokens
• Cache read: $0.015 / 1M tokens

You're welcome to start using it. Actual billing follows the pricing page.

✅ Upstream recovered: OpenAI, Claude, and Grok APIs are available again

The widespread outage in the North American server clusters of OpenAI, Anthropic (Claude), and xAI (Grok) has been resolved. All model APIs from these three providers are available again.

If you still see isolated request failures, please retry shortly. Thank you for your patience.

⚠️ Major upstream outage: all OpenAI, Claude, and Grok APIs are currently unavailable

Due to a widespread outage in the North American server clusters of OpenAI, Anthropic (Claude), and xAI (Grok), all model APIs from these three providers are currently unavailable.

This is an upstream provider outage, not a Modelflare platform failure. We are monitoring recovery and will update this notice as soon as service is restored.

Please pause requests to OpenAI, Claude, and Grok models for now. We apologize for the inconvenience.