🐳 DeepSeek V3.1 Debuts With Big Upgrades
DeepSeek has rolled out V3.1, a hybrid model that toggles between reasoning and standard reply modes, showing clear gains in agent skills and tool use and even slightly edging past R1-0528 on key benchmarks.
➡️ The update is now live in the web app—simply click DeepThink to enable reasoning.
Read More
Post #597
123K
- ❤ 166
- 👍 28
- 🔥 28
- 🥰 10
- ❤🔥 9
- 💘 8
- 🙏 7
- 🗿 6
- 😱 1
- 👾 1