Hi everyone! The Chinese startup DeepSeek has released V3.1: a huge update to its AI lineup.
The model now uses reasoning and handles complex tasks, coding, math, and science questions much better than the base version.
Tests show the new model is on par with Claude Sonnet 4. In some cases, it even outperforms GPT-5 mini.
You can already use DeepSeek V3.1 completely free in our bot, in both normal and thinking modes.
How to Use?
🔴 Go to @GPT4Telegrambot
✅ Select DeepSeek-V3.1 (basic mode) or V3.1 Thinking (reasoning mode) in the /model section or with the Choose model button.