try-working

5 ms·

61.
by try-working 2mo ago
yeah it's pretty cool. you can also use it as a regular music player. if you toggle autoplay, it will automatically keep playing the next track in your track list. supports uploading local libraries and youtube, and recording, uploadin
62.
by try-working 2mo ago
I recently wrote about first principles of model routing that I've learnt building a model router. The model pool should be kept small, and models in the pool should be clearly differentiated. For example, one large frontier model for
63.
by try-working 2mo ago
I built something similar, but for the browser: https://dj.t-tunes.com/
64.
by try-working 2mo ago
Now let's see Dario's price cut.
65.
by try-working 2mo ago
pi-role-model
66.
by try-working 2mo ago
OpenAI is doing this so that you can't preserve cache when routing between for example GPT 5.6 and DeepSeek V4, or other providers. I'm building a router that easily reaches >99% cache hit rate in Pi when routing between GPT 5.
67.
by try-working 2mo ago
Let's see how the market reacts.
68.
by try-working 2mo ago
DeepSeek themselves called it `deepseek/deepseek-v4-flash`. Pro is still like that.
69.
by try-working 2mo ago
Agree on 5.4. And it has 4x more quota than 5.6.
70.
by try-working 2mo ago
combine Codex with Cursor. both are really good. no other subs are worth it.
71.
by try-working 2mo ago
i've never had any issues with 256k context. see no reason to bump up to 1m if it comes at a premium.
72.
by try-working 2mo ago
This is just superstition.
73.
by try-working 2mo ago
Anthropic must be forced to open source all of the books they have scanned and burned.
74.
by try-working 2mo ago
I've been using Cursor on auto for the past five days while waiting for my Codex refresh. Yes, it works extremely well. Extremely.
75.
by try-working 2mo ago
Cursor is probably a better client than Claude Code though
76.
by try-working 2mo ago
he's pausing because of the leak. the transcript content has nothing to do with it.
77.
by try-working 2mo ago
I have the second largest Kimi plan, the Chinese version. When K2.6 was their latest model, the quota was good; it was like GPT $100 is now or what the $20 version was in December. When K2.7 was released, they cut quota by 80%. I can't
78.
by try-working 3mo ago
another option for on-device routers is role-model that i'm building: https://github.com/try-works/role-model
79.
by try-working 3mo ago
It's not China. It's individual labs. Open source is their go to market strategy: https://try.works/why-chinese-ai-labs-went-open-and-will-rem...
80.
by try-working 3mo ago
No, open models is a marketing strategy
81.
by try-working 3mo ago
This is their motivation https://try.works/why-chinese-ai-labs-went-open-and-will-rem...
82.
by try-working 3mo ago
sounds like a caching issue, and maybe other issues too
83.
by try-working 3mo ago
everyone is using OpenAI and Anthropic in China. We have both providers at work as well.
84.
by try-working 3mo ago
in case someone wants to speculate in why chinese labs open source their models: https://try.works/why-chinese-ai-labs-went-open-and-will-rem...
85.
by try-working 3mo ago
input lag is a minor part of it. it depends on the game, but for example in CS2, refresh rate changes game mechanics like recoil, movement, even trajectories of thrown grenades. i went from 60hz to 240hz, with <100fps average, and the di
86.
by try-working 3mo ago
it's recursion
87.
by try-working 3mo ago
no, you download the runtime and connect your models to it, then you use the router as the endpoint in your coding agents
88.
by try-working 3mo ago
you could use my model router to route between models like that. https://github.com/try-works/role-model
89.
by try-working 3mo ago
It's a GTM strategy: https://try.works/why-chinese-ai-labs-went-open-and-will-rem...
90.
by try-working 3mo ago
He was made to leave.

More ›