DeepSeek
Frontier-level reasoning at a fraction of the cost, and open weights.
Overview
DeepSeek shook the industry by showing that models rivaling the Western frontier could be trained and served for a tiny fraction of the assumed cost, then releasing the weights in the open. For developers, that combination of quality, price, and self-hostability is close to irresistible.
Its reasoning models expose their chain of thought, making them a favorite for anyone who wants to see how an answer was reached.
What it is great at
Value is the headline. DeepSeek delivers top-tier math, coding, and reasoning at API prices that undercut the majors by an order of magnitude, and you can run it on your own infrastructure to avoid per-token costs. The transparent reasoning traces are a genuine differentiator for technical work.
Where it falls short
DeepSeek is a China-based company, and its default hosting raises data-governance questions for regulated industries and privacy-focused users. Self-hosting sidesteps this but requires resources. The model also applies content restrictions on sensitive political subjects, and the consumer app trails the polish of ChatGPT or Gemini.
Pros & cons
What we like
- Reasoning and coding that rivals far pricier models
- Open weights you can inspect, fine-tune, and self-host
- Far lower cost than Western frontier labs
- Transparent chain-of-thought on hard problems
What could be better
- China-based hosting raises data-governance concerns for some
- Content restrictions on sensitive political topics
- Consumer app is less polished than the majors
The verdict
The disruptor that proved frontier reasoning need not be expensive. A superb value, with the usual caveats about where your data goes.