Would I be surprised there’s bench maxing happening? Yes. But some users also use Q4 quantized and complain how dumb local models are.

I am so happy right now, qwen3.6-27b was an absolute game changer. To see another one in the same league.. phew

Does anyone have a https://tenstorrent.com/hardware/cards to try it on?

Model benchmarks are useful, to a point, but it is the long tail of things you do with the model that determines if it's good at a wide range of activities. Ant/OAI, to their credit, build their models -- even the small ones -- so they follow instructions and do tool calling well, without the system prompts confusing them. This is especially important for long-horizon tool calling.

So one open weight model might "meet" Opus or whatever on benchmarks, but then fail to follow a simple answer format and also tool call correctly. The models are whipped to within an inch of their lives to strictly adhere to their post training quality gates.

Man what a week, We just had GLM 5.3 that came out and then we had smaller local model Qwen3.8-27B from Qwen

Just tried using Pi Agent and looks very promising

I'm wondering whether any provider can offer this for cheaper $/token than the new DSv4 Flash, which is both cheaper and smarter :/

Completely local use is a different story, of course.

Need to upgrade to a second 3090! Slowly building up my local models with Krea2, MiniMax H3 (and their new Music3), and now Qwen 3.8

Really excited to see what people do with this. 3.7 27B was probably the best compromise between size and intelligence to run on consumer hardware

I wish this can run directly on my RTX 4090, seems like 30B is the sweet spot for dense model to run locally, sadly RTX 5090 is very expensive and I need a new PC and new power supply(and UPS) to run that, adding a second RTX 4090 is another option, but not sure if my PC can do that yet.