Today, DeepSeek released an update to its V3 model. Called V3.1, the new model features a longer context window. Beyond that, the market is still digesting what the new model has on offer. Early reviews are too nascent to judge, and no major benchmarking has been released.

Why do we care? DeepSeek’s open-source models are very popular, snagging 15% of recent inference requests on OpenRouter’s platform — ahead of OpenAI. Thus, if V3.1 is good, it could once again raise doubts that the closed-source approach that most Western AI companies are taking is mistaken. Comparisons to OpenAI’s two recently released open models will prove irresistible.

https://huggingface.co/deepseek-ai/DeepSeek-V3.1-Base/discussions?_bhlid=d97fd199ba168f23bcf9db0eb67ae94b7f914278

 #🇨🇳 ChinA.I. 🤖🧠🦾🤖

deepseek-ai/DeepSeek-V3.1-Base · Discussions

🇨🇳 ChinA.I. 🤖🧠🦾🤖

Today, DeepSeek released an update to its V3 model. Called V3.1, the new model features a longer context window. Beyond that, the market is still digesting what the new model has on offer. Early reviews are too nascent to judge, and no major benchmarking has been released.  Why do we care? DeepSeek’s open-source models are very popular, snagging 15% of recent inference requests on OpenRouter’s platform — ahead of OpenAI. Thus, if V3.1 is good, it could once again raise doubts that the closed-source approach that most Western AI companies are taking is mistaken. Comparisons to OpenAI’s two recently released open models will prove irresistible.  https://huggingface.co/deepseek-ai/DeepSeek-V3.1-Base/discussions?_bhlid=d97fd199ba168f23bcf9db0eb67ae94b7f914278   #🇨🇳 ChinA.I. 🤖🧠🦾🤖

Jerome Cody

A group of likeminded people in 🇨🇳 ChinA.I. 🤖🧠🦾🤖 are talking about this.