431–440 of 525 posts
I couldn't get past the first chart which basically showed that intelligence and cost are both better for gpt luna, I'm not sure what the argument is here for Gemini flash 3.7 given that comparison.
not what the first chart shows
Earlier quoted context omitted.
Think of the potential layoffs of highly paid employees! But, I think it’s also based on what they are being used for, most LLM users are still mainly SWEs or similar as I understand and there’s a ton of data to train them for coding.
I agree its what they are being used for and their primary revenue source. My point is mainly that was never the pitch that got ai the hype it did and imo doesn't justify the valuations even if we all lose our jobs to ai. Because it no longer seems like they even think its making other jobs go away.
Same with those that use to improve marketing outcomes for SEO and such, now there are AI based reports with automatic improvements.
Finally on other domains you already have robots on supermarkets, fast food, and gas stations, where the customer does the work of the (now gone) employees without any kind of price reduction.
Earlier quoted context omitted.
Matched roughly with Sol on DeepSwe cost per task. Luna way cheaper. DeepSeek used to be, but I think it's somewhere on Sol's curve after the price hike.
On DeepSwe it's strictly beaten by Luna on max, cost and result. Damn, Luna on max is as good on DeepSWE as Kimi k3, I think I dismissed this model unjustly.
Have you tried the new DeepSeek Pro v4, Qwen 3.8, Gemini 3.7 Flash, and Grok 4.6? Do they make sense for any use cases? I'm currently using omp with Kimi K3 as the planner and DeepSeek v4 Flash 0731 as the implementer, or CC + Fable for planning and Opus 4.8 for implementation. For API(not coding), I just use DeepSeek v4 flash 0731 and MiMo. I'm pretty happy where I am, but I'm wondering if these new models provide s…
Just use all of them and synthesize final result, record scoring when synthesizing. Later you can make decision to drop low performing ones.
Earlier quoted context omitted.
what do you mean by reset at every turn? context stays until compaction. if you remove the reasoning tokens after every turn you will be constantly blowing cache which is far worse than filling up context.
That's not my understanding of how most agents work. This is what a chain of request/response looks like: Your Prompt 1: Prompt Content 1 -> cache-1 LLM Response 1: Thinking Content 1 Response Content 1 Your Prompt 2 (client side): prompt-1 + response-without-thinking-1 + Prompt Content 2 Your Prompt 2 (server side): cache-1 + response-without-thinking-1 + Prompt Content 2 -> cache-2 LLM Response 2: Thinking Content…
Earlier quoted context omitted.
Just use all of them and synthesize final result, record scoring when synthesizing. Later you can make decision to drop low performing ones.
Synthetize=?
Earlier quoted context omitted.
gemini flash is probably the best model for visual tasks right now. they also make it really easy to ingest videos
Crazy it's still the only video understanding endpoint. It's what I use it for and no other model even offers a competitor.
Earlier quoted context omitted.
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...
Fable orchestrating DeepSeek v4 Flash to implement a plan is my new favorite thing. It's so freaking fast, but you gotta tell Fable to watch Deepseek like a hawk or it'll go off the rails.
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 lead on price. It's like 5x cheaper than 3.6 and well 2.5x cheaper than 3.7 "introductory price". Comparable quality.
Earlier quoted context omitted.
not that I'm intending to defend Sergey here but, if you could spend 0.04% of your net worth to protect 5% of your net worth (and probably all of your easy liquidity)... wouldn't you?
I think that argument should stop with advocacy for law. I have no problem with this when it's means playing optimally within the current laws, things like hiring world class tax attorneys. But if using your immense wealth to ensure the rules that everyone plays by favor you specifically, instead of the country/world, It feels like this is closer to bribing a judge than hiring a tax attorney.
A one-off retroactive wealth tax is a rather questionable form of taxation even if (like myself) you are in favour of looking for ways to tax the super rich more effectively.
When I read about these one-off wealth taxes I find one thing really astonishing. The argument against wealth taxes in general is that people would change their behaviour, which would reduce the tax take over time. That's certainly true.
But then left wing economists believe that imposing it retroactively and calling it a "one-off" does not change people's behaviour. Technically, it cannot change their behaviour with respect to this specific taxation event.
But I find it utterly naive to believe that this sort of hit and run taxation will not change people's behaviour in other ways that would reduce tax revenues.
No one in their right mind would ever believe that a tax that brings in $100bn over 5 years will not have to be replaced by some other tax paid by the same group or by the somewhat less well off who decided to stay.