Let’s compare with the deepseek version again. From this point of view, it is still more refined than the ds 👀
Is Qwen really going to stand up this time? Then 27B is more worth looking forward to🤓☝️ In addition, Teacher Loktar’s prompts are really well written, and they are also worth learning.
Wow!!!!!!!!! The level of detail of Qwen4 is even better than that of deepseek V4.1 flash 1017…
Wow!!!!!!!!! The level of detail of Qwen4 is even better than that of deepseek V4.1 flash 1017 version!!! The key is that it is more efficient and saves tokens!!! Teacher Loktar’s Bee Sunflower Qwen4 version!!! If the release of qwen4 really has this effect, it would be really good!!!!!!! My expectations for qwen4 27b are getting higher!!!!!
How to connect Codex to DeepSeek? Switch between DeepSeek and GPT with one account, and CC Switch can change models with just one click.
After reviewing Musk's interview with CCTV Finance in September, I found that his evaluation of China's AI is much higher than I thought.
Lao Ma said that the Chinese model performed very well under limited computing power. If measured by the performance of unit computing power output, China may be the strongest in the world! ! Even more boldly, he predicts that China is expected to solve the computing power bottleneck caused by lithography and chip manufacturing within 2-3 years! ! !
You have reached the end of the archive
All of deepseek