Telegram agents stuck on ChatGPT OAuth rate limits – need correct model/provider switching
✅Solved
I need help @Krill 🦐 validating the correct way to handle provider switching.
Current state (working but blocked):
• Clawdbot running locally on macOS (gateway reachable, config parses)
• Telegram channels enabled and responding (with rate limit notice)
• Multiple long-running Telegram group sessions in progress (cannot delete or reset)
• Sessions were originally created using ChatGPT OAuth (openai-codex / gpt-5.2)
Problem:
• ChatGPT Plus usage limit is reached for the day
• Telegram replies now fail with OAuth rate-limit messages
• Even after updating agents.defaults.model and restarting the gateway, existing Telegram sessions continue using OAuth instead of switching to API
• Bot is “up” but effectively unusable until OAuth resets
Config highlights:
• OAuth profile still enabled (openai-codex:default)
• OpenAI API + Gemini API available as backups
• Telegram currently not switching providers automatically
Current state (working but blocked):
• Clawdbot running locally on macOS (gateway reachable, config parses)
• Telegram channels enabled and responding (with rate limit notice)
• Multiple long-running Telegram group sessions in progress (cannot delete or reset)
• Sessions were originally created using ChatGPT OAuth (openai-codex / gpt-5.2)
Problem:
• ChatGPT Plus usage limit is reached for the day
• Telegram replies now fail with OAuth rate-limit messages
• Even after updating agents.defaults.model and restarting the gateway, existing Telegram sessions continue using OAuth instead of switching to API
• Bot is “up” but effectively unusable until OAuth resets
Config highlights:
• OAuth profile still enabled (openai-codex:default)
• OpenAI API + Gemini API available as backups
• Telegram currently not switching providers automatically
Solution
@Krill 🦐 Okay, took quite a bit of troubleshooting but the solution was actually quite simple. I was able to type /model openai/gpt-5.2 directly into Telegram and it was as simple as that... and it switched models.
So as long as you have your API keys and the models are set up, if you hit rate limits or get stuck you should literally just be able to type in /model provider/model and it will instantly switch it lol. Idk why that was so difficult to figure out and why there's no documentation on that anywhere.
So as long as you have your API keys and the models are set up, if you hit rate limits or get stuck you should literally just be able to type in /model provider/model and it will instantly switch it lol. Idk why that was so difficult to figure out and why there's no documentation on that anywhere.
The AI that actually does things. Emails, calendar, home automation - all from your favorite chat app. New shell, ready to help. The lobster way.
83,815Members
Resources
Was this page helpful?
