2 ms·
Look for Deepseek 4 when it drops, I’m curious how good it will be. The thing is, if you’re using AI responsibly today you’re already breaking down tasks to su
by nothinkjustai 7mo ago
Look for Deepseek 4 when it drops, I’m curious how good it will be.
The thing is, if you’re using AI responsibly today you’re already breaking down tasks to such a granular level that you don’t need the power of Opus. You can save that for deeper research tasks.
- storus 7mo agoBased on the current DeepSeek website I suspect it's not going to be great as their current model (V3.4? V4-mini?) often forgets or changes facts explicitly mentioned in the conversation which R1 never did. It's better than R1 at math or coding, but nearly unusable for deep conversation. I suspect they pushed MLA or linear attention too much, or quantize a lot more than before.