Guppy is an easy-to-use AI chat client that supports OpenAI, Deepseek, Anthropic, and Google AI models. Store and restore conversations effortlessly. Use your own API key.
Settings
Model:
Max Tokens:
Temperature:
Top P:
Presence Penalty:
Frequency Penalty:
Reasoning Effort:
Dark Mode: Markdown:
Dark Mode: Markdown:
Configuration
- As of now it only supports text input and output, but other media are considered.
- You use your own API Key, so that your account is charged for the use.
- If you don't have an API Key, register one at OpenAI Platform, Deepseek Platform, Anthropic or Google Gemini, and create one there.
- Hover over settings and buttons for a more thorough description.
- O1 and O3 models don't support most of the settings for now, as per OpenAI, and only O1 and O3 Mini support Reasoning Effort.
- Streaming and System are supported for all models except those from Anthropic. This will be further investigated.
URL Arguments
Optionally use these to save the look-n-feel configuration as a bookmark.
- apikey sets your API Key for a certain provider. Don't share a URL containing your API Key!
- model sets the AI model. Use the token for the model as specified by the provider.
- darkmode and markdown can be used for respective option. Set to 1 or true to activate.
- parser sets what markdown-to-HTML parser to use. Options are marked (default) or showdown.
Roles
- System: Optional and can be used to set the behavior of the assistant. Use first in the conversation. Can't be used with O1 models.
- User: Provides requests or comments for the assistant to respond to.
- Assistant: The assistant's response, but can also be written by you to give examples of desired behavior.
Download/Upload
- Downloaded or externally created JSON, YAML and CSV files can be uploaded to continue a conversation.
- Only JSON and YAML will include model settings, so preferably use those formats for archiving and continuing conversations.
- Text, HTML and RTF are intended for use by other applications and can't be uploaded to Guppy.
Advice
- Write your prompts as instructions rather than as questions.
- Be as descriptive and clear as possible about what you expect as a response and be careful to spell correctly to avoid confusion.
- Use significant keywords and phrases consistently so it's clear how different aspects relate.
- Specify constraints by setting limits on length, format, or style.
- Define the response format by telling the model how to structure it.
- Iterate the conversation to clarify and correct.
Privacy
- Nothing is stored centrally about API Key, prompts or responses, or anything else for that matter.
- We only collect anonymous usage statistics at loading of the page via our own Matomo installation.
