DeepSeek review
Open-weight reasoning that rivals GPT-5
4.7
freemium20M+ usersTop RatedTrendingCommunity TrustedFree Tier
90
ToolHund Score
Elite
Top 5%
Our verdict
DeepSeek's R1 and V3 models match top closed labs on math and code — and the weights are open. Free web chat plus a cheap API.
ToolHund Score: 90/100 · Best for: Power users wanting frontier reasoning for free.
Strengths
- R1 reasoning model
- Strong fit for power users wanting frontier reasoning for free.
- Real free plan you can ship work with
Weaknesses
- Best features sit behind the paid plan
- Output quality varies with prompt skill
- Limited offline use — internet required
Pricing
Free · API from $0.14/M
Strong — free plan covers most casual use, paid unlocks pro workflows
Features
- R1 reasoning model
- Open weights
- Free web chat
- Very low API cost