maxdo

6 ms·

121.
by maxdo 4mo ago
Their digital sovereignty
122.
by maxdo 4mo ago
curious how many people model killed on battlefields of ukraine and russia.
123.
by maxdo 4mo ago
running so much compute on the scale is not a junior task. weird analogy
124.
by maxdo 4mo ago
going to IPO is a sign of confidence , you need to report a lot of things, that private companies don't. This is an exact reason chinese labs do not rush to go public. They wish to go , but money flow that is not as good. On the same n
125.
by maxdo 4mo ago
I see exactly opposite . Chinese models fails under any complex scenarios, while us labs raise the price , that's a sign of confidence.
126.
by maxdo 4mo ago
Majority of my agentic setup is pi / Claude code where every single Chinese models are not as good except commercial 1T models . Local is a pipe dream . If you can run it cheap occasionally why commercial companies can’t run it cheaper
127.
by maxdo 4mo ago
Why on earth I should switch from a top tier model to much worse local model ? Why do I need to suffer my battery ?
128.
by maxdo 4mo ago
It’s a vertical company they did compute very good , their top model bounce between top tier and -1 -2 gen. They were top tier only once though on paper briefly . If tomorrow thy will hit top tier , that do have know how to expand . They ca
129.
by maxdo 4mo ago
i tried glm 4.7 for agents that write code. simple scripts 200-1000 LOC. extremely bad . Had to abandon cerebras oferning, their smart models are only on enterprise plan.
130.
by maxdo 4mo ago
Both companies created in the previous century and running on old fuel of past success
131.
by maxdo 4mo ago
You just named symptoms not cause. 1. People vote for old idiots . 2. Local politics is all about milking any potential work with permits. Sky rocket housing due to that . 3. American lost engineering culture everywhere except software . No
132.
by maxdo 4mo ago
Lantern | Software Engineers (Python, Full-Stack) | NYC (Onsite) | Full-time | withlantern.com Lantern is building AI-powered revenue intelligence—autonomous agents that turn cold outreach into warm conversations. We detect buying signals a
133.
by maxdo 4mo ago
Great UI cult could be an early pre-ai phenomena , not a lost decade. you will need 100 times less of front end , if any action can be asked or explain in voice, text or video or slides always custom to your request. you don't need nav
134.
by maxdo 4mo ago
Oh most prominent eu ai company . Without reading an article predict next, will update after : 1. They give up on building competitive models. It’s time to drink wine not to struggle with competition 2. Because of #1 they will talk a bit ab
135.
by maxdo 4mo ago
they use a "low" flavor of it to scale it on billions of users
136.
by maxdo 4mo ago
benchmarks we deserve: google search quick ai answers vs full llm model :)
137.
by maxdo 4mo ago
it's not that huge of a deal if you compare commercial costs in china and cheapest us states, and electricity is only one of the factors. The real reason: anthropic + openai just cut the reasoning output to prevent distill, and hence y
138.
by maxdo 4mo ago
a flagship with no pirates, all fired due to ai.
139.
by maxdo 5mo ago
this price only available if you ok to send your data to Beijing Volcano Engine Technology Co. for the rest open router vendors it is not the same.
140.
by maxdo 5mo ago
Flash on it's own is not a very competitive model, it's pricing is within ranges of everything else on the market. Probably the most direct competitor of Flash model : GPT 5.4 mini Cache Read $0.075 /M tokens Gemini 3 flash :
141.
by maxdo 5mo ago
GPT 5.4 Cache Read ≤272K $0.25 And it's multi modal, and available at whatever you might imagine rates limits.
142.
by maxdo 5mo ago
Sonnet : Cache Read $0.30 Gemini 3.5 flash : Cache Read $0.15
143.
by maxdo 5mo ago
maybe i need to give it second chance, surprisingly Kimi 2.6 consistently fail even to generate valid json plan, where gemma 4 was doing really good, but slow.
144.
by maxdo 5mo ago
i've been trying that, in reality every time you try to save it, it's not worth it, the cost of mistake is so high , you can spent 2-3h on just wrong assumption, you lost your time and all the burned tokens.
145.
by maxdo 5mo ago
I'm not curious what tasks you tested it for. Im working on coding agent writing code dynamically on request for customers. i'd say code itself very simple and aggressively cached, and patternalized, e.g. we adding lots of hints t
146.
by maxdo 5mo ago
Populist , let’s invent Make intelligence great again :)
147.
by maxdo 5mo ago
No opus 4.7 , gpt5.5 , Gemini flash 3.5 in benchmarks
148.
by maxdo 5mo ago
chinese ai is not doing the same? or they don't parse? they do except they also send thousands of sex-spies to do espionage of this kind on the scale.
149.
by maxdo 5mo ago
sure, sounds like QA lol Scale: Over 3.4 million exchanges The operation targeted: Agentic reasoning and tool use Coding and data analysis Computer-use agent development Computer vision Moonshot (Kimi models) employed hundreds of fraudulent
150.
by maxdo 5mo ago
How can distilled opus become better than original? There are numbers of reports including anthropic that kimi team was participating in fraudulent activities

More ›