341–350 of 481 posts
Earlier quoted context omitted.
Yeah either the benchmark isn't very useful anymore or V4 Flash is a really, really good model.
GPT 5.6 Luna is an extremely cheap and still very capable model. A chinese model being in the same ballpark of capability at half the price sounds believable to me.
I've been using it extensively since the release and the best summary I can give is that it's good enough to use it for (almost) everything and cheap enough that the cost are irrelevant. I'm running it in Oh My Pi with a second instance running as "advisor" and even with 5-6 active sessions (effectively 12 streams) I'm struggling to spend more than 5 bucks per day. OpenCode Go even has double limits temporarily so fo…
Another time, Flash started trying to make tool calls by just calling bash and catting the tool call to stdout. Then it started running echo xx for every two letter UNIX command it could think of: mv, cp, etc and the it dug into uv, ty, and jj
And how this has been accelerating!!
I felt this very hard when I had to travel in the middle of nowhere in south america, with no network, and wanted to keep an LLM model on my macbook pro with 48GB of RAM. That was back in April 2026, a few months ago.
I downloaded Google Gemma 4 (google/gemma-4-26b-a4b) and - Oh boy - I was amazed by it's capacity!
I was able to use it to code simple things, ask it about nature, learn new stuff while traveling and make stories for the kids.
Was really amazing to observe and experiment this!
Seems to me there will be some good chance to run these great LLM locally on our hardware!
Amazing time to be alive
Earlier quoted context omitted.
I think you're overlooking the fact that for long-horizon tasks, even small errors compound over time and can lead to catastrophic outcomes. For simple queries, we have reached the threshold since the beginning of the year, and models are good enough from every provider to make a meaningful difference between one another. (ChatGPT, Claude, Gemini, Grok, MuseSpark, Kimi, DeepSeek, GLM...) The real unlock will be, and…
I have a silly (but honest) question. What's an example or two of a > 24hr task that people are actually asking something to do? Like real life ones.
Another one I did was a printer data stream translator from an obscure format to PostScript/PDF (or just PNGs), complete with cups support, etc so these old apps can easily be hooked up.
Flash is capable now of running long range defined-goal tasks like this.
Earlier quoted context omitted.
From opencode go $10/mo plan I get between 60 t/s and 100 token/s even with large contexts of 150k+ tokens. I wouldn't call 80 t/s slow.
You are right, relatively to other llm providers this is not slow. But if you think what is possible when you have 1000t/s a sec you might find it slow.
Compared to the last Deepseek V4 Flash version I've had tons of issues with it getting in infinite loops and talking to itself without executing tool calls, wasting tons of tokens This is on Pi agent, nothing fancy at all about my prompts or use case. Anyone else experiencing this? I've also had it randomly go from talking about Rust to talking about the electric chair, controversies about D&D rules (both irrelevant…
What quantization are you using? Which infra provider? Baseten.co's version got into a loop rather rapidly... I've since added loop detection and adjusted some other settings on the pi coding agent and have yet to notice it again. I also switched to DeepInfra ... who serves an fp4 version admittedly, but I've had no issues with it as of yet and it's the top provider on openrouter.ai volume wise.
Earlier quoted context omitted.
not at all true. if you're truly using it across the board for smaller things (translation of pages, filtering of every individual tweet based on its relevance to you etc), the costs ramp up super quickly. i used for work where i did less and it quickly reaches thousands if you're not careful. i can already see what some will say: skill issue et cetera - whatever.
I thought it went without saying that GPT 5.6 Sol is the wrong model to use for things like filtering tweets. Apparently not?
Earlier quoted context omitted.
Not counting the cost of making the model, which is subsidized by… someone? The chinese gov i think?
DS comes out (one of, or) the most successful quant fund in China. They don't strictly need any kind of subsidies. FWIW they have a funding round planned (kerfuffle about leaks from CEO presentation few weeks back) -- presumably because infrastructure needs have ballooned. Naturally there will be some PRC government interest in one of their flagship AI companies. From what is visible seems to be more along the lines…
> They don't strictly need any kind of subsidies.
You understand how these two sentences directly contradict each-other, yeah? The money-losing operating of training a model is paid for by momey earned from prior investments. So… the work is “subsidized” by its parent company’s investments in it.
Earlier quoted context omitted.
And what do you do with all that?
I've reverse engineered multiple classic games and turned them into popular, browser-based MMO-like experiences. I'm also creating a free platform that replaces extremely out-of-date software, some of it only available with mutli-million dollar contracts, to help medical physics professionals with cutting-edge radiotherapy devices used to treat cancer. https://brynnbateman.com/ for a list of projects
And out of curiosity, how do you automate testing the porting in the browser that's actually playable etc? And aren't you a bit scared of hosting and serving the "hairy bits" such as full assets? Nice job anyway!
Earlier quoted context omitted.
Not counting the cost of making the model, which is subsidized by… someone? The chinese gov i think?
Subsidized by inference profits and volume.