All models
DeepSeek
DeepSeek V3.1
DeepSeek's August 2025 hybrid that switches between quick answers and step-by-step thinking in one model. A capable, cheap coder built on a 671B open-weight mixture-of-experts, text only. DeepSeek V4.1 Flash is newer at roughly a third of the price.
OpenOpen-weight models with lighter filters. Discuss more, refuse less.
Smart & Cheap
Intelligence
~0.71 credits
Estimated price per message
164K tokens
Context window
Text
Reads
DeepSeek V3.1 is a legacy model. Its successor, DeepSeek V4.1 Flash, is faster or cheaper - and chats that used DeepSeek V3.1 move to it automatically.
See DeepSeek V4.1 FlashThe facts
Made by
DeepSeek
Content tier
OpenOpen-weight models with lighter filters. Discuss more, refuse less.
Intelligence
Smart & Cheap
Estimated price per message
~0.71 credits≈ 140 messages for 1 €
Context window
164K tokens≈ 252 pages of text in one go
Reads
Text
Writes
Text
Tool use
Yes - search, memory, code and more
Served via
OpenRouter, Unbottled AI
Status
Legacy
A typical message: 16K tokens in, 1.5K out. Short replies cost less. Models served via OpenRouter bill the real provider cost plus our markup, so actual charges vary. 1 credit = 0.01 €.
Compare DeepSeek V3.1
Same facts, side by side.
Try DeepSeek V3.1 now.
No subscription. Free credits to start. The chat opens with DeepSeek V3.1 selected and a prompt ready - just send.