Uploaded August 2026 | Updated September 2026, 3 hours ago
Open vs. closed models. Once more and for all đ€
Dont get me wrong, we need open source models. Itâs amazing for research and progress.
Itâs just not a necessity for AI engineers.
Open weight models are a fantastic category for most companies, even the big ones like Kimi K3.
Sure, most of us wonât host these ones locally. But, through various API providers, they provide a valuable pricing war competition to OpenAI and Anthropic beneficial for the consumer.
I highly recommend trying out deepseek-v4-flash for example as a very cheap powerful alternative to Gemini and other good models your companies might be running on. #short
Open vs. closed models. Once more and for all đ€
Dont get me wrong, we need open source models. Itâs amazing for research and progress.
Itâs just not a necessity for AI engineers.
Open weight models are a fantastic category for most companies, even the big ones like Kimi K3.
Sure, most of us wonât host these ones locally. But, through various API providers, they provide a valuable pricing war competition to OpenAI and Anthropic beneficial for the consumer.
I highly recommend trying out deepseek-v4-flash for example as a very cheap powerful alternative to Gemini and other good models your companies might be running on. #short
