blagui

11 ms·

1. 1.▲
   by blagui 1mo ago

   You have 4 thinking levels. You can disable it. It's well known issue in Qwen, previous releases I would disable it by default. Also xhigh seem a new thing.
2. 2.▲
   by blagui 2mo ago

   A 9B beating frontiers" No! But nailing specialized tasks Yes.
3. 3.▲
   by blagui 2mo ago

   GLM is the model that will sink the frontier labs. Recall last year deepseek? And 18 month's later? What changed?
4. 4.▲
   by blagui 3mo ago

   Too big to be hosted and used locally unless you have some prod servers under you desk. And those aiming to fit with Q2 or Q1. It's not even worth it to destroy the models to claim it's still alive after cutting all the limbs.
5. 5.▲
   by blagui 3mo ago

   So the sweet spot for dev in 2026 is 64k context windows? Are we back in 2024? As more context will degrade a lot the t/s. On top this is 1 slot. If you use sub agents the kv cache will be invalidated with colliding request and make it
6. 6.▲
   by blagui 3mo ago

   How you can do dev in 2026 using 64k context and without sub agents? The benchmark seemed fine until I saw that. If you use sub agents, they will overwrite the cache and each request will trigger full reprocessing. Have fun with that as it