DeepSeek
DeepSeek · General AI · Coding · Research
Chinese AI lab whose open-weight reasoning models drew global attention. The chat app is capable and free; developers can run the models anywhere.
Best for: Strong open-weight reasoning models at low/no cost.
| Models | DeepSeek model family (open weights) |
|---|---|
| Free plan | Yes |
| Paid plans | API is paid; chat largely free |
| API | Yes |
| Custom instructions | No persistent instruction box. Paste per conversation |
| Memory | No cross-chat memory on the standard app |
| File upload | Yes |
| Web search | Yes |
| Context window | Model-dependent |
| Apps | Web · iOS · Android |
Facts last reviewed 2026-08-10, capabilities and plans change frequently; verify on the official site before relying on a detail.
Models, limits and pricing
| Current models | DeepSeek-V4-Pro, flagship reasoning model (deepseek-v4-pro) DeepSeek-V4-Flash, fast, high-concurrency (deepseek-v4-flash, 0731 build) Both released as open weights under the MIT license |
|---|---|
| Context window | 1M tokens; 384k max output (both V4 models) |
| Parameters | V4-Pro 1.6T total / 49B active; V4-Flash 284B total / 13B active |
| Modalities | Text in and out; thinking mode; tool calls and JSON output |
| Consumer pricing |
|
| API pricing | deepseek-v4-pro: $0.435 / 1M input (cache miss), $0.87 / 1M output |
Pricing and model line-ups verified 2026-08-11 from source 1 ↗ · source 2 ↗ · source 3 ↗. These change often. The vendor page is authoritative.
Notable
- Open weights, self-hostable
- Visible chain-of-thought reasoning mode