DeepSeek
Open-weights frontier models from China; chat product + API
Overall score
About
DeepSeek is a Chinese AI lab that releases competitive open-weights frontier models, including the DeepSeek-V3 family and the reasoning-focused DeepSeek-R1. Operates a free consumer chat product and a low-cost API. R1 became famous in early 2025 for matching frontier reasoning at a fraction of the inference cost.
Best for: Cost-sensitive builders who want frontier reasoning at a fraction of OpenAI / Anthropic API prices, and willing to accept Chinese data-residency posture.
Pricing
Chat (Free)
- Monthly
- n/a
- Annual /mo
- Free
- Billing
- flat
- Notes
- Free consumer chat product;DeepThink reasoning mode;File upload
API
- Monthly
- n/a
- Annual /mo
- n/a
- Billing
- usage
- Notes
- Pay-per-token API access;DeepSeek-V3 + R1;Off-peak discounts · Per-million-token pricing is the lowest of any frontier-class model.
| Tier | Monthly | Annual /mo | Billing | Notes |
|---|---|---|---|---|
| Chat (Free) | n/a | Free | flat | Free consumer chat product;DeepThink reasoning mode;File upload |
| API | n/a | n/a | usage | Pay-per-token API access;DeepSeek-V3 + R1;Off-peak discounts · Per-million-token pricing is the lowest of any frontier-class model. |
Key features
- Free consumer chat
- DeepSeek-R1 reasoning model
- Open weights (MIT licence on R1 distillations)
- Aggressive API pricing
- Strong on math + code
Integrations
Trust & compliance
- Stage range
- n/a
- Founded
- 2023
- Status
- active
- SOC 2
- no
- GDPR
- unknown
- Data residency
- cn
- External rating
- n/a
- Last verified
- May 2026
Reviews
Be the first to share your experience.
Related tools in LLM chatbot
Pairs well with
People who've discussed DeepSeek
See all people →Curated mentions from podcasts, posts, and public stacks. Editorial coverage; not endorsements.
Clem Delangue
Co-founder & CEO
Clem Delangue publicly endorses DeepSeek as a canonical example of an open-weights frontier model worth training-against. In his Ultra-Scale Playbook X announcement he positions DeepSeek R1 alongside Llama and GPT-5 as the proof points that every tech company should be able to train their own frontier model. He's been a vocal amplifier for the China-led open-source AI thesis on X and LinkedIn.
Dharmesh Shah
Co-founder & CTO
Dharmesh Shah surfaces DeepSeek as one of the model options inside Agent.ai (alongside o3-mini, Claude 3.5 Sonnet, Flux and Ideogram). He's been deliberate about exposing open-weights models like DeepSeek to agent builders rather than locking them into a single frontier vendor, which signals his read on where the open-source model layer is heading.
Eric Schmidt
Former CEO
Eric Schmidt includes DeepSeek alongside ChatGPT, Gemini, Claude and Grok in the set of frontier-model AI tools he sees in widespread daily use. The Chinese open-weights model gets equivalence-class treatment from him, consistent with his broader US-China AI competition commentary.
Marc Andreessen
Co-founder & General Partner
Marc Andreessen publicly called DeepSeek's release 'AI's Sputnik moment' in January 2025, treating the open-weights Chinese frontier model as a structural shift in the AI competitive landscape. He frames open-weights models like DeepSeek as a hedge against closed-API lock-in and an existence proof that frontier capability isn't gated to a single vendor.
