Loading...
Product
Core Features
Top Extensions
Company
Subscribe to our newsletter.
Get product updates and news in your inbox. No spam.
Straight from your command bar, ask anything that you wanted and get an AI-generated answer without any effort.
Customize the model to your liking. Create and edit custom engines beyond your creativity.
Continue talking about everything right where you left off. Be pro without from zero.
Got the answer that you wanted? Great. Now you can save it without asking again.
Automatically save all the question and answer so you can go back digging for the answer you're looking, quickly.
Process text taken from anywhere (selected text, clipboard text, opened web page) and insert the result into the frontmost application or copy it to the clipboard.
Windows 11+ is supported. For vision commands, selected images are read from the active File Explorer window or from the clipboard (both platforms).
Manage chat presets and AI commands together in Models. Select a preset and use Create AI Command from This Model in its action menu to create a command with the same chat settings.
AI Commands is available for quick execution and command management. The standalone Create AI Command entry has been removed; use the action menu in Models to create commands. Existing commands are preserved. Shortcuts or deep links to the removed entry need to be replaced with Models.
GPT-5 model supports vision capabilities, which can be enabled in the Models Command when creating or editing a model.
You can also enable per-model reasoning control in the model form and set the Effort (none, low, medium, high).
By default, reasoning effort override is disabled. When it is enabled and set to anything except none, the extension sends reasoning_effort in Chat Completions requests.
Enable Use API Endpoint and set API Endpoint in preferences to use a compatible API provider.
When creating or editing a Model or AI Command, the Model dropdown loads model IDs from the configured API.
Search and select an available model, or type any model ID and select Use "your-model-id" to use it.
Manual entry is always available, including when the API cannot list models. There is no separate Custom model preference.
Previously saved model IDs remain selectable even if they are missing from the API response.
Azure continues to skip model discovery; enter the model ID manually.
This extension requires a valid Secret Key as your API Key from OpenAI with a pay-as-you-go plan account (you'll get a 429 error if you're on a free-tier account).
All the preferences value will be stored locally using Preferences API
All preferences properties list that can be customize through Raycast Settings > Extensions > ChatGPT
| Properties | Label | Value | Required | Default | Description |
|---|---|---|---|---|---|
apiKey | API Key | string | true | empty | Your personal OpenAI API key |
useStream | Stream Completion | boolean | true | true | Stream the completions of the generated answer |
isAutoSaveConversation | Auto-save Conversation | boolean | true | true | Auto-save every conversation that you had with the model |
isHistoryPaused | Pause History | boolean | false | false | Pause the history of the conversation |
isAutoLoadText | Auto-load | boolean | false | false | Load selected text from your frontmost application to the question bar or full text input form automatically |
isAutoFullInput | Use Full Text Input | boolean | false | false | Switch to full text input form from question bar automatically whenever you want to ask or type a question |
isAutoTTS | Text-to-Speech | boolean | false | false | Enable auto text-to-speech everytime you get a generated answer |
useApiEndpoint | Use API Endpoint | boolean | false | false | Change the OpenAI's default API endpoint to custom endpoint |
apiEndpoint | API Endpoint | string | false | empty | Custom API endpoint |
useProxy | Use Proxy | boolean | false | false | Each question request will be passed through the proxy |
proxyProtocol | Proxy Protocol | http, https, socks4, socks5 | false | http | Proxy protocol option |
proxyHost | Proxy Host | string | false | empty | Proxy host value |
proxyUsername | Proxy Username | string | false | empty | Proxy username value |
proxyPassword | Proxy Password | string | false | empty | Proxy password value |
useAzure | Use Azure OpenAI | boolean | true | false | Use Azure OPENAI rather than OPENAI |
azureEndpoint | Azure Endpoint | string | false | empty | Azure OpenAI resource endpoint |
azureDeploymentName | Azure Deployment | string | false | empty | Azure OpenAI resource deployment |
Copy and paste your Azure OpenAI's KEY value to the API key field
Copy and paste your Azure OpenAI Endpoint value to the Azure Endpoint field. Then, Tick the Use Azure OpenAI checkbox
Copy and paste your Azure OpenAI Model deployment name value to the Azure Deployment field
Donate to support the development of this extension. Thank you!
Made with ♥ from Indonesia
Collection of your saved answer
Ask ChatGPT via Raycast
Collection of your recent answer
Collection of your recent conversation
Collection of your custom and default model
Summarize Website and YouTube video
Template with a plain detail view
Template with a plain detail view
Run and manage AI commands
Command: prefix. Saved conversations retain their settings if the command is deleted.Custom model preference; saved model IDs remain available when model discovery fails or omits them.Reasoning toggle with an Effort selector (none, low, medium, high) in the model form.reasoning_effort in Chat Completions requests only when the reasoning override is enabled.reasoning_effort when OpenAI-compatible providers do not support that parameter.gpt-4o-mini and the like to gpt-5-nano, as the latter is cheaper, more capable and more efficient.useModel initialization process to update localStorage only after initialization is complete, preventing data overwrite. Addresses issues #8977, #14356, and #14125.List.EmptyView when the model list is empty, as the Action panel would remain empty and default models ensure the model list is not empty.gpt-3.5 and the like to gpt-4o-mini, as the latter is cheaper and more efficient.setOption to include all model IDs, removing the filter for 'gpt' prefix. (To support o1 models).src/type.ts:Model.option, since 'option' was already effectively a 'string'.isAutoFullInput state machine checkisAutoTTS not available in stream modeList.Item.Detail.Markdown display big image caused Heap out of memorySummarize command to summarize website and YouTube videolocalstorage to cacheModels (Enable in preferences)Auto-load Text and Use Full Text Input not working properlyAsk onSelectionChange race conditionPreferences dynamically directly from the generated type definition429 error codeAPI -> APIKey (@k8scat)429 error codeIntroducing ChatGPT for Raycast. Interact with OpenAI's ChatGPT straight from your command bar, with 5 commands available:
Ask Start a new conversation with ChatGPT by asking a questionConversations → Continue recent conversations without losing contextModels → Collection of the user's default and custom modelsSaved Answer → Collection of the user's saved question and its answerHistory→ Collection of all the user's questions plus the answers