DeepSeek
by DeepSeek
Text Generation
Freemium
✨ Builders and budget operators who want frontier-grade reasoning at API prices that round to zero.
The Chinese open-weight lab that embarrassed the industry: frontier-class chat free at chat.deepseek.com, with an API priced in fractions of cents.
About DeepSeek
DeepSeek's V4 family — V4 Flash for speed, V4 Pro for heavy reasoning — delivers near-frontier performance under an MIT license, with the consumer chat completely free and the API costing roughly $0.14 per million input tokens on Flash (Pro a few times more). V4 Pro is a 1.6-trillion-parameter mixture-of-experts with a 1-million-token context window and serious coding scores (SWE-Bench Verified 80.6%). Most people do not know the chat product runs two brains: pick Instant for quick answers or Expert for the full reasoning model. The trade-offs are real: avoid client-confidential material on the free chat, and expect politically sensitive topics to be filtered. Creator use-case: process an entire feature script's worth of notes in one 1M-token context window, then draft the shoot-day call sheet for free.
Key Features
-
⚡V4 Flash: fast and cheap — roughly $0.14 per million input tokens
-
⚡V4 Pro: 1.6T-parameter mixture-of-experts, 49B active, built for reasoning
-
⚡1M-token context window with 384K max output
-
⚡Chat modes: Instant for quick answers, Expert for deep reasoning
-
⚡Open weights under MIT license — self-hostable
-
⚡Free consumer chat at chat.deepseek.com, no paid subscription tier
Pros
-
✓Free chat with genuinely frontier-class output
-
✓API pricing undercuts Western labs by an order of magnitude
-
✓Open weights: run it yourself, no vendor lock-in
Cons
-
✕Free chat is not for confidential or client-sensitive material
-
✕Politically sensitive topics are filtered
-
✕Western compliance certifications and SLAs lag OpenAI and Anthropic
Suggested Alternatives
📚 Continue Learning
Deepen your skills with related courses, prompts, and research within the AINewZ ecosystem.