pimeys

5 ms·

31.
by pimeys 2mo ago
Yep. It's a bit scary also. There's a lot of opportunity in the market now, but the downfall of the big US inference labs is going to hurt here in EU too sadly...
32.
by pimeys 2mo ago
It's interesting if they need to cut off their subscriptions to be able to compete in API prices. Very interesting...
33.
by pimeys 2mo ago
They just raised their prices sadly.
34.
by pimeys 2mo ago
Internal reports from company? Maybe not. I'm just saying you have to eval eval eval if you are working in this industry. There's a ton of victories in price, and price is right now the key thing all the customers are talking abou
35.
by pimeys 2mo ago
Long-context agentic tasks and Rust engineering are our use cases where Kimi definitely is better than Sol. We can measure our own systems and the numbers say that Sol has no chance against K3 or Opus, and K3 is so so so much cheaper than O
36.
by pimeys 2mo ago
And Fireworks did not yet. They are still under the limit of not feasible to self host... Let's see if other providers follow DeepSeek with their flash pricing.
37.
by pimeys 2mo ago
The competition is real in pricing. Thanks for the Chinese open models, US big players have to cut their inference pricing. We've done a bunch of evals between the models, and Kimi K3 was the first one that actually could compete or be
38.
by pimeys 2mo ago
I bought a OnePlus 6 for 50 euros from eBay and I can do a ton of things with it already. Plasma is great in mobile, and what did not work I got working pretty fast together with Kimi and DeepSeek. For example here's patches to get NFC
39.
by pimeys 2mo ago
Not only Lyme... In the seaside Finland the tick-borne encephalitis cases are going up. And that's no joke either, but at least you get a vaccine for that... if you remember to take it three times before the ticks wake up in springtime
40.
by pimeys 2mo ago
And a few of his other series, such as Treme, The Deuce, and Generation Kill are all fantastic.
41.
by pimeys 2mo ago
If you are working in a company and using language models, it is a very good idea to hold a bunch of evals you can trust and use to validate new models. Calibrate every once in a while with prod data. We have our own and the only numbers on
42.
by pimeys 2mo ago
Not the parent, but: https://omp.sh/ You define roles for different agents like this: modelRoles: task: fireworks/kimi-k3-fast:high plan: fireworks/kimi-k3-fast:max slow: fireworks/kimi-k3-fa
43.
by pimeys 2mo ago
Yes. It works very well for simple tasks. When I know the context grows over 200k, I implement with Kimi. We run an agent company and we do a bunch of different things with agents. Where we used Gemini before Deepseek v4 Flash is taking the
44.
by pimeys 2mo ago
It is also very good and cheap for computer use.
45.
by pimeys 2mo ago
Yes it is cheap, but per task DeepSeek v4 Flash is a bit more expensive and lands between Terra and Gemini 3.6 Flash in quality. Closer to Gemini than Terra...
46.
by pimeys 2mo ago
Hey me too. This is my morning Zen moment refactoring and cleaning some code before I go back to the agent to earn money. One of the only things that helps me to relax is to do some manual coding...
47.
by pimeys 2mo ago
But if it's 30-50% cheaper than Opus, provides same or better results and you can deploy it to your own premises or choose one from many providers and pay per token? It is, in general, measurably cheaper per task than Opus is.
48.
by pimeys 2mo ago
They are profitable and active on the enterprise local model territory. You can RL a model with them for your own purposes and I heard good things about it.
49.
by pimeys 2mo ago
Yeah, been running too many evals in the past week I start to mix the versions up. Probably should sleep... We run an agent company and outside coding the new Gemini 3.6 Flash and GPT 5.6 Luna are very interesting. Luna can do a bit of rese
50.
by pimeys 2mo ago
If you have agents and users, you can run evals and see how far the models go. Luna is not greatest in tool calls, but if you define your problem well and the tools well, it is comparable to Gemini 4 Flash with much lower price tag.
51.
by pimeys 2mo ago
I don't remember can you swim in Lietzensee but you definitely can in Weissensee and I consider both to be quite central still. But you are correct, there's not enough swimmable lakes in Berlin and when the weather is nice you bet
52.
by pimeys 2mo ago
Yes. I tested OpenClaw that burned 15€ just by starting it and I hated its configuration. Spent 20€ in tokens and built my own in a day, that does everything I want and sips tokens. What a time to be alive.
53.
by pimeys 2mo ago
Funny. I use Fable a lot at work, but this one I paid from my own pocket and coded it with GLM 5.2. Paid 20 euros in total.
54.
by pimeys 2mo ago
I have these around the house: https://www.home-assistant.io/voice-pe/ Then you can set the background AI to be any OpenAI compatible API. So I just created one to my local Rust Agent, and connected it to home assistan
55.
by pimeys 2mo ago
I wrote my own OpenClaw one weekend and I am running it as my assistant through Matrix with DeepSeek v4 Flash (and Qwen). It probably costs me about 2 dollars a month and is even more useful than ChatGPT would be due to me having full contr
56.
by pimeys 2mo ago
I've used GLM-5.2 a lot on fireworks and had never ever issues on rate limits. If they cannot handle the load with K3, there's the priority tier to get your evals done. I'm definitely having full eval suite on already if they
57.
by pimeys 2mo ago
Depends what you do. We have certain tasks we spend money on where Gemini 4.6 definitely is better than Opus 5.
58.
by pimeys 2mo ago
I remember when 4.7 and 4.8 were released and people were asking what's wrong with them and 4.6 is the best. But yes, I also think it's not the greatest model for programming. On the other hand, for agentic tasks that are not prog
59.
by pimeys 3mo ago
Depending on the harness, $200/session can be common. Thanks for harnesses like maki the cost is lower compared to OpenCode or Cursor. But $200/day is very common in the business. The monthly plans are just a trial version of what
60.
by pimeys 3mo ago
Ty

More ›