Hy-4 versus Deepseek Flash.
Using Deepseek Harness. I still really love the Deepseek output but Hy4 put up a great showing with about a quarter of the tokens and a 3rd of the requests.
I edited this video in T3 Code using OpenCode with DeepSeek V4.1 flash
What did you think? 👀 Returning to posting content on Instagram
Everyone says the chinese models have caught up.
So I had the big 3 race Astra in my sailing sim Kimi struggled while costing 2x Astra. GLM, Deepseek barely made it out of the blocks. the frontier models are just so much better. cheap tokens ≠ cheap agents
4 Harbor demos, 3 ran locally one via API.
I really am getting beginning to get antsy regarding Qwen 4 release, I NEED to see how well it's going to do. All of these feel good in their own rights but Maya EXL3 2.9bpw is on an entirely different level every time, I don't get it
You have reached the end of the archive
All of deepseek