Codex error
Selected model is at capacity. Please try a different model.
Codex prints this as written. It is a capacity message, not a usage limit.
The model is full right now.
It is not your usage limit.
What it means
The model you picked is out of capacity on the vendor side. Your plan usage is not involved.
Codex treats it as retryable. It waits for the retry time the server gives, or backs off with jitter.
Fix it
- Pick a different model in Codex, as the message says.
- Or wait a few minutes and run the turn again.
Switch while you wait
If a model stays full for a long time, run the same task on a vendor you already pay for.
Or point Codex at your own OpenAI API key and a model you choose.
With superbot
superbot is your own AI agent with a model of its own that runs on superbot credits, so it keeps working when a vendor's plan runs out.
It also puts your own API keys behind one base URL, with one caller key, in the OpenAI or Anthropic format.
Calls on your own keys spend no superbot credits, and you pay your provider as usual.
Codex takes an OpenAI base URL, so see the OPENAI_BASE_URL page.
Related
Source: openai/codex: codex-rs/protocol/src/error.rs. Read on 2026-10-17.