Use a DeepSeek API key with CalorieCodex’s Custom OpenAI-compatible provider.
Enter your DeepSeek settings
- Create an API key on the DeepSeek Platform.
- In CalorieCodex, open Settings → AI quick log and tap Set Up or Manage.
- Under Provider, select Custom OpenAI-compatible.
- Paste your DeepSeek key into API key (optional). DeepSeek requires a key even though this field is optional for some compatible endpoints.
- Set Base URL to
https://api.deepseek.comand Model ID todeepseek-flash. CalorieCodex adds the chat-completions path automatically. - Tap Save or Save Changes. Start with a short text estimate and review the result before logging it.
Choose a setup for photos and Agent features
Image input and tool calling depend on the model and endpoint. If a conversation fails after an Agent tool call, DeepSeek’s thinking mode can require reasoning data in the follow-up request. For that workflow, switch to a compatible model through OpenRouter or use Gemini.
See DeepSeek API docs for current model IDs and DeepSeek thinking mode for tool-call requirements.
Keep your API key safe and limit spending
If you use an API key, set up these safeguards in your provider’s dashboard.
- Use a separate key for CalorieCodex. Give it a name you’ll recognize, and keep it out of other apps, screenshots, and support messages. If it’s exposed, revoke it and create a replacement.
- Set a low spending limit. Use a limit on the key if your provider offers one. If limits are managed at the project level, check whether the setting blocks requests or sends an alert.
- Start with a small prepaid balance. If your provider supports prepaid credits, start with $1–$5 where allowed, or the provider’s minimum. Leave automatic reload off and add credits manually when you need them.