FRONTIER LAB / LLM · WORTH A LOOK
DeepSeek
Chinese AI lab with open-weight LLMs (DeepSeek V4 family) and a low-cost API.
What DeepSeek does
Chinese AI lab with open-weight LLMs (DeepSeek V4 family) and a low-cost API. Open-weight frontier LLMs plus a chat app and a pay-per-token API with off-peak discounts and search/fetch tools.
Good for
- Very low-cost LLM calls for automations
- Open-weight model for self-hosting
Try this in 10 minutes
- Very low-cost LLM calls for automations. Pick one real example from your own business and try just that.
- Open-weight model for self-hosting. Put the result next to how you do it today. Is it faster, cheaper or better?
- Sign up (free plan or trial if available) and try it on one real task from this week.
Mini-tasks based on common small-business uses for this tool.
Honest take
No red flags in our notes so far
- Check data-residency requirements before client use.
- Major model provider.
Pricing
API pay-per-token, e.g. input (cache miss) $0.15–$1.32/M and output $0.6–$3.96/M depending on model and peak/off-peak.
Prices as listed by the maker when we last checked (2026-10-20); usually in US dollars. Check the website for current prices.
Latest update
- 2026-10-20: Open-sourced DeepInfer inference framework (powers V4.1-Flash) and deepseek-recipe.
- : Open-sourced DeepInfer, the in-house inference framework that serves DeepSeek-V4.1-Flash (optimised kernels, distributed inference, speculative decoding, structured outputs), plus deepseek-recipe for API-matching token encoding/decoding, so teams can self-host with the same behaviour as the official API at lower compute cost.
- : US DHS order made public: federal civilian agencies must stop buying and remove DeepSeek products, including third-party products built on DeepSeek models.
- : DeepSeek-V4.1-Flash-1017 released (company says better benchmarks, faster, shorter answers); Search & Fetch APIs opened to all at $0.75 per 1,000 calls each, free inside DeepSeek Harness, Claude Code and Codex until 31 Oct 2026.
Is DeepSeek worth it? Quick answers
Is DeepSeek worth it?
DeepSeek is worth a look if it matches a job you already do (Frontier lab / LLM). Check data-residency requirements before client use. It is a paid tool, so test it on one real task before committing.
How much does DeepSeek cost?
API pay-per-token, e.g. input (cache miss) $0.15–$1.32/M and output $0.6–$3.96/M depending on model and peak/off-peak.
What is DeepSeek used for?
Chinese AI lab with open-weight LLMs (DeepSeek V4 family) and a low-cost API. Typical uses: very low-cost LLM calls for automations; open-weight model for self-hosting.