AJ(@ItsmeAjayKV):More flash models please !! Flash models are getting bigger unfortunately, but it's very clear they are getting much more capable. Deepseek-v4.1-flash is better than v4.0-pro. Their focus on inference efficiency matters. Faster, cheaper inference benefits local users, and even when a model is too big for our hardware, cloud providers can use that efficiency to serve more people with the same compute. That means more people can get access to capable open models. This is also why i'm excited for Qwen4-flash. More intelligence, more efficiency, but not more size. Give us more flash.

2026.10.09 13:00

More flash models please !! Flash models are getting bigger unfortunately, but it's very clear they are getting much more capable. Deepseek-v4.1-flash is better than v4.0-pro. Their focus on inference efficiency matters. Faster, cheaper inference benefits local users, and even when a model is too big for our hardware, cloud providers can use that efficiency to serve more people with the same compute. That means more people can get access to capable open models. This is also why i'm excited for Qwen4-flash. More intelligence, more efficiency, but not more size. Give us more flash.