🚀 Claude Opus 5 is now live for all groups!

You can switch to claude-opus-5 with the following command:
claude --model claude-opus-5

Or execute it directly inside Claude Code:
/model claude-opus-5

💰 Pricing: Exactly the same as Opus 4.8. Enjoy the upgrade!

🚀 Kimi k3, MiniMax M3, & GLM 5.2 are now LIVE!
🛡 100% Official Channels: Direct & genuine APIs (No third-party self-hosted instances).

🔥 Grok is 75% off right now

📌 Welcome to the MetaRtr Channel
Affordable, reliable AI API access through a simple, unified interface.
🔹 OpenAI-Compatible API
Simply change your Base URL and integrate with your existing applications.
🔹 Reliable Infrastructure
Designed for stable API access and consistent performance.
🔹 Pay-As-You-Go
Transparent pricing with no hidden fees.
💰 Competitive Pricing
Save up to:
• 90% on Grok
• 87% on OpenAI
• 80% on Claude
🌐 Get Started
https://www.metartr.com
💬 Support
@MetaRtrSupport_bot
MetaRtr is an independent third-party AI API service and is not affiliated with OpenAI, Anthropic, xAI, or other model providers.

📢 Official Price Drop Alert for ChatGPT Models!
We have updated our prices in line with the latest official price reductions:
📉 gpt-5.6-terra: Price reduced by 20% 📉 gpt-5.6-luna: Price reduced by 80%

🔗 Base URL [https://api.metartr.com/v1]
🎨 Image Endpoint [https://api.metartr.com/v1/images/generations]

📢 Notice: OpenAI Official Rate Limit & Overload Issues
You may encounter the following official OpenAI error messages due to high upstream server traffic:
🚨 Selected model is at capacity. Please try a different model.
🚨 Our servers are currently overloaded. Please try again later.
📌 Please note: These are exact error messages generated directly by OpenAI’s official servers. This issue is entirely on OpenAI's end and is not caused by our platform load, your account status, or local network settings.
Currently, this capacity constraint mostly affects GPT-5.6 models. To avoid interruption, we recommend temporarily switching to GPT-5.5 or waiting until OpenAI's official capacity recovers.

📢 Model Maintenance Notice

Please note that the gpt-5.6-luna model (Codex group) will be temporarily taken offline starting August 6, 2026, at 02:00 UTC. Access will be unavailable until further notice.

⚠️ Action Required: If your workflows rely on this model, please migrate to an alternative model or adjust your configurations promptly to avoid service disruption.

For any questions or assistance, feel free to contact Support.

— MetaRtr Team

🚀 Reverse Proxy Access for Google Gemini is Now Live-75% OFF! Special Intro Pricing for Testing.

GPT-5.6 Sol Frequent Reconnections During Peak Hours
Due to a recent surge in global inference demand on OpenAI’s side, GPT-5.6 Sol may experience model capacity overload during peak hours.
Previously, the system would show: Selected model is at capacity. Please try a different model.
We have now implemented an automatic retry mechanism. As a result, the client will display: Reconnecting... 1/5 (or the Chinese equivalent)
Please note: This is not an API disconnection or an issue on our servers. It is triggered by capacity overload on OpenAI’s upstream model, which activates the automatic retry process.
If you encounter this message frequently during peak hours, we recommend temporarily switching to GPT-5.5.
— MetaRtr Team

📌 Claude Group Adjustment Notice
Due to insufficient AWS upstream resources, Claude lite (Sale) will be taken down today at 13:00 (SGT, UTC+8).

🔹 Recommended Action
Lite users are advised to switch to Claude Plus (Premium) as a priority.

🔹 Automatic Migration
To prevent service disruption, users still on the lite tier after 13:00 will be automatically switched to Plus—no manual action needed.

🔹 Risk Warning & Alternatives
Recent AWS risk control measures have tightened, which may cause stability fluctuations for Claude Plus (Premium). If you experience frequent errors, we recommend switching to the CCMAX group or other available models.

🙏 Support & Understanding
Thank you for your understanding and support!

📢 Pricing Update & Peak/Off-Peak Billing Announcement

DeepSeek pricing has been officially adjusted as of August 17, 2026, at 00:00 (SGT / UTC+8), introducing dynamic peak and off-peak billing.

Affected Models:
• deepseek-v4-flash
• deepseek-v4-pro

Time Slots & Rates (SGT):
⚡️ Peak Hours: 09:00 – 12:00 | 14:00 – 18:00 (Standard Rate)
🌙 Off-Peak Hours: All other times (50% OFF peak price)

Thank you for your continued understanding and support!

【Fix for 401 Error After Codex 0.149.0 Upgrade】

If you encounter API_KEY_REQUIRED / 401 Unauthorized after updating, it is because the new version no longer allows custom providers to automatically inherit authentication from auth.json when requires_openai_auth=false.

Please modify your configuration file:

* macOS: ~/.codex/config.toml
* Windows: %USERPROFILE%\.codex\config.toml

Ensure your configuration is set to:

model_provider = "codex"
model = "gpt-5.6-sol"

[model_providers.codex]
name = "codex"
base_url = "https://api.metartr.com/v1"
wire_api = "responses"
requires_openai_auth = true
supports_websockets = false

And retain auth.json:

{
  "OPENAI_API_KEY": "YOUR_METARTR_API_KEY"
}

Save and restart Codex. You can also re-run the latest install script to apply the fix automatically.

Note: Do not rely on auth.json while keeping requires_openai_auth=false; this is the root cause of the 401 error in 0.149.0.