Error in OpenAi Response - Engine not found

[DylanSimowitz]

Describe the bug

OpenAI error in API response engine not found.

To Reproduce Steps to reproduce the behavior: e.g.

1. Select some text
2. Press <Leader>mcs

This is using the default setup copied directly from the readme. No changes to configuration at all.

Expected behavior

Return result from OpenAI

Error

{"status": 404, "headers": ["date: Fri, 22 Oct 2021 22:39:23 GMT", "content-type: application/json", "content-length: 129", "access-control-allow-origin: *", "openai-organization: user-bnsxbna6vrmqknqvlmfcgwnb", "openai-processing-ms: 12", "openai-version: 2020-10-01", "strict-transport-security: max-age=15724800; includeSubDomains", "x-request-id: 9136d7e345981c1dc78e25227d659c76", "", ""], "exit": 0, "body": "{\n  \"error\": {\n    \"code\": null,\n    \"message\": \"Engine not found\",\n    \"param\": null,\n    \"type\": \"invalid_request_error\"\n  }\n}"}

Details (please complete the following information):

• OS: MacOS
• Neovim version: 0.5.1
• nvim-magic version: latest master

[jameshiew]

Thanks for the report, does your OpenAI account have access to use the davinci-codex engine? That "Engine not found" error looks like it might be returned in the case where you don't - https://github.com/mattt/OpenAI/issues/2 . If you only have access to the davinci engine you could substitute that backend in the plugin setup, though the responses from the API won't be as good.

require('nvim-magic').setup({
	backends = {
		default = require('nvim-magic-openai').new({
			api_endpoint = 'https://api.openai.com/v1/engines/davinci/completions',
		}),
	},
	use_default_keymap = false
})

NB: the davinci engine isn't free to use like the current private Codex beta so this could result in charges per call made.

[joshbode]

Thanks - this worked for me using the Codex model code-davinci-002, which is in limited beta:

api_endpoint = "https://api.openai.com/v1/engines/code-davinci-002/completions"