DeepSeek
Chinese open-weight reasoning model — strong at maths and code, runs locally or via API.
Add DeepSeek to your hut →DeepSeek is a Chinese AI lab whose open-weight models — DeepSeek-R1 (reasoning) and DeepSeek-V3 (general) — benchmarked near Claude and GPT-4o on maths and coding tasks at dramatically lower cost. The web chat is free; the API is priced well below OpenAI equivalents, making it popular for developers building cost-sensitive applications.
The main considerations are data privacy (servers in China) and occasional content refusals on politically sensitive topics. For pure technical performance per dollar, few options compete.
| Made by | DeepSeek |
|---|---|
| Pricing | Free (web chat) · API (very competitive pricing) |
| Best for | Maths, coding, technical reasoning, cost-efficient API |
Alternatives to DeepSeek
- Claude
Anthropic's long-context assistant — strong reasoning, coding, and nuanced writing.
- ChatGPT
OpenAI's flagship assistant — broad capability, plugins, image input, and GPT-4o.
- Gemini
Google's assistant with deep Search integration, 1M token context, and Workspace hooks.
- Grok
xAI's assistant with real-time X/Twitter data access and image generation via Aurora.
- Copilot
Microsoft's AI assistant — integrated with Bing, Office 365, and Windows.
- Kimi
Moonshot AI's frontier assistant — K2.6 model, 1M token context, agentic "OK Computer" mode for slides and web tasks.