401 invalid_request_error you didn't provide an api key

[jangid]

I am using the following format in ~/.authinfo

machine api.openai.com login apikey password sk-secret-openai-api-key-goes-here

but still get the error - 401 invalid_request_error you didn't provide an api key

what could be wrong?

[jangid]

Additional info:

It works fine if I use with default configuration. But when I setup like this it gives the error.

(declare-function gptel-make-ollama "gptel-ollama")
(declare-function gptel-make-anthropic "gptel-anthropic")
(declare-function gptel-make-gemini "gptel-gemini")
(declare-function gptel-make-openai "gptel-openai")
(use-package gptel
  :config
  (defvar gptel-backend-anthropic (gptel-make-anthropic "Claude" :stream t :models '(claude-3-5-sonnet-20240620 claude-3-7-sonnet-20250219)))
  (defvar  gptel-backend-ollama (gptel-make-ollama "Ollama" :stream t :host "localhost:11434" :models '(deepseek-coder-v2)))
  (defvar  gptel-backend-gemini (gptel-make-gemini "Gemini" :stream t :models '(gemini-2.5-pro-exp-03-25)))
  (defvar  gptel-backend-openai (gptel-make-openai "ChatGPT" :stream t :models '(gpt-4o gpt-4.5-preview)))
  (setq gptel-model 'gpt-4o
	gptel-backend gptel-backend-openai)
 )

[karthink]

(defvar gptel-backend-openai
  (gptel-make-openai "ChatGPT"
    :stream t
    :models '(gpt-4o gpt-4.5-preview)
    :key gptel-api-key))

See the documentation of gptel-make-openai.

Any reason you don't want to use the built-in OpenAI backend?

[jangid]

Thanks.

I want the list of models to be small in the gptel-menu, so wanted to customize that.

[karthink]

Makes sense.

You could also just set the list of models for the default backend:

(setf (gptel-backend-models (gptel-get-backend "ChatGPT"))
      '(gpt-4o gpt-4.5-preview))