DeepSeek

DeepSeek offers a consumer AI assistant and a separate developer API. The official app is free to chat with; the website also links to web chat. API calls are billed by tokens, model, cache status and time period, so a free consumer app does not imply free API access.

✨ Key Features

• Consumer chat — The official app lists Deep-Think, web search, uploaded-file text extraction and cross-platform chat history. Available controls and service limits may vary.
• Developer models — The current API table lists deepseek-flash (DeepSeek-V4.1-Flash) and deepseek-v4-pro (DeepSeek-V4-Pro-0813), each supporting thinking and non-thinking modes.
• API context and output — Both list a 1M-token context and maximum 384K output; these are API model limits, not promises of consumer-chat quotas.
• Structured integrations — JSON output, tool calling and supported API compatibility options are listed in the API documentation.
• Vision boundary — Flash supports vision; the listed V4-Pro API does not. Do not assume identical multimodal features across models.

💳 Pricing

Consumer app chat is free; the developer API charges USD per one million tokens:

Peak periods are 01:00–04:00 and 06:00–10:00 UTC on Monday–Friday, excluding Chinese public holidays. Other times, weekends and Chinese public holidays are off-peak. API fees are deducted from granted or topped-up balances, using granted balances first when both exist. Prices and model limits can change; no fixed free API allowance is promised here.

⚖️ Alternatives Compared

🎯 Use Cases

• Discussing ideas or extracting text from a supported file in the app.
• Using optional search and reasoning controls for a research question.
• Building an application with structured output or tool calls.
• Selecting an API model according to vision requirements and token costs.