A.I.
DeepSeek
DeepSeek’s current API line is V4: call deepseek-flash (DeepSeek-V4.1-Flash) or deepseek-v4-pro. Older names deepseek-chat and deepseek-reasoner were retired after 24 Jul 2026 UTC. R1 and V3 stay as archive doors.
| Model | Status | Price | Released |
|---|---|---|---|
| FlashCurrent recommended API model. DeepSeek-V4.1-Flash.deepseek-flash · 1M · lead card | Current | $0.15 in, $0.60 out (off-peak, cache miss) | 10 Sep 2026 (announced) |
| V4 ProStill listed on Models & Pricing. Service continues after 14 Sep 2026 (changelog).deepseek-v4-pro · DeepSeek-V4-Pro-0813 | Current | $0.66 in, $1.98 out (off-peak, cache miss) | Not on file |
| ThinkingDual thinking / non-thinking mode, shared across Flash and Pro.Default on · effort low / high / max | Feature, a setting on Flash and V4 Pro | See page | Not on file |
| APIBase URLs, current model ids, and retired aliases.api.deepseek.com · OpenAI and Anthropic formats | Feature, endpoints and model ids | See page | Not on file |
History
| Model | Status | Price | Released |
|---|---|---|---|
| R1Landmark open reasoner (Jan 2025). Not the current API SKU.Archive stub | Replaced by the V4 line (Flash and V4 Pro) | See page | 20 Jan 2025 |
| V3Prior chat generation. API name deepseek-chat is retired.Archive stub | Replaced by the V4 line (Flash and V4 Pro) | See page | Not on file |
Prices are US dollars per million tokens (word pieces) in and out unless noted. "Not on file" means this site has no checked value yet.