Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Gemini 3.5 Flash

blog.google

Post reply on HN
Print view View original

441–450 of 692 posts

Re: Gemini 3.5 Flash

#441

Post by karmakaze »

simonw wrote: post #131

The pelican is a lot : https://github.com/simonw/llm-gemini/issues/133#issuecomment... Not a great bicycle though, it forgot the bar between the pedals and the back wheel and weirdly tangled the other bars. Expensive too - that pelican cost 13 cents: https://www.llm-prices.com/#it=11&ot=14403&sel=gemini-3.5-fl...

I'm hoping we'll have many of these pelican cyclist pictures collected. Then when all the models can do it well, we'll stop posting about them, and dhen the next generations of AIs train on the data we'll have these canonical archetypes.

Re: Gemini 3.5 Flash

#442

Post by dsdsfaa »

mynameisash wrote: post #371
overrun11 wrote: post #357

Earlier quoted context omitted.

Yes that is exactly what is happening. OpenAI and Anthropic are the fastest growing companies by revenue ever and their gross profit margins are healthy.

According to this article[0]: > HSBC Global Investment Research projects that OpenAI still won’t be profitable by 2030, even though its consumer base will grow by that point to comprise some 44% of the world’s adult population (up from 10% in 2025). Beyond that, it will need at least another $207 billion of compute to keep up with its growth plans. This article is from six months ago. Was HSBC wrong; did something dr…

and to make matters worse, they are massively over-valued.

Whoever buys the stock at a richly priced 1tn at ipo is a bozo lmao. I know I know, index funds will be forced to hold it bypassing the 1 year rule. Disaster already.

Re: Gemini 3.5 Flash

#443

Post by hedgehog »

SwellJoe wrote: post #428
sheeshkebab wrote: post #417

Earlier quoted context omitted.

You can only run heavily quantized models on all 3/4/5 rtx gpus (with 32gb or less vram) - and you probably want moe versions like Qwen 35b for this to run at speed somewhat comparable to Claude. It’s still not there to be honest but getting there. Personally I mess around with llama.cpp on m5 max with 128gb - it’s a decent setup to try various medium sized things, and runs llms surprisingly well without quantization…

Two 3090s is 48GB, so it's possible to run the 6-bit quantization comfortably, which is fine. It doesn't start to get notably dumber until lower than that. It won't be as fast as a hosted model, but dual 3090s will be comfortably fast for interactive use with the MoE version and not terrible to use with the dense model. I run the dense model at 8 bits on my dual Radeon V620 desktop machine, which I think would be slo…

Have you done comparisons with 4 bit and seen a noticeable difference for coding tasks?

Re: Gemini 3.5 Flash

#444

Post by hedora »

margorczynski wrote: post #331

Wow at the price hike. Still I think in the long run the Chinese will win if they're able to produce hardware comparable to Nvidia.

Why would the Chinese sell me nvidia cards? I can just by an AMD iGPU, and the perf/$ is much better than nvidia dGPUs.

(Typed on a 2023 macbook perfectly capable of running the Chinese open weight models.)

Re: Gemini 3.5 Flash

#445

Post by superchink »

jonnyasmar wrote: post #385

The $1.50/$9.00 pricing is a meaningful shift if you've been running Gemini as the "fast iteration" half of a multi-model coding workflow. I've had Claude Code, Codex, and Gemini CLI running side by side and the working split was "Gemini for quick scaffolding and exploration where the cost of being wrong is low, Sonnet for correctness-critical stuff." At 3x the Flash pricing that split stops making sense — you're pay…

Out of curiosity, what was your workflow to generate this comment? I’m curious what model (claude?) and process (manual prompt with bullet points?) you used.

Re: Gemini 3.5 Flash

#446

Post by dsdsfaa »

overrun11 wrote: post #353
roadside_picnic wrote: post #199

Earlier quoted context omitted.

These companies are unprofitable (as all companies at this stage and ambition should be) but I increasingly don't see any justification for the idea that it is fundamentally unprofitable. Inference alone is certainly profitable. I'm running models at home that are comparable to performance of paid models a year or so ago for free. Even for much larger models the cost around inference serving are clearly manageable. T…

Arguably nothing even has to change with training for this to be sustainable. Dario has claimed that Anthropic is profitable on a per training run basis. They aren't profitable because they choose to keep investing in increasingly large training runs.

Cut the crap.

The value of the firm's operating assets = EBIT(1-t) - Reinvestment

You (Anthropic) want that sky-high valuation? Accept reinvestment is part of the equation.

If they decide to stop reinvesting, then they are as good as dead.

Moreover, they clearly are not re-investing cash flows from operations. Why do you think they are continually raising money? Lmao.

Re: Gemini 3.5 Flash

#447

Post by rohansood15 »

TurdF3rguson wrote: post #280
spyckie2 wrote: post #222

Earlier quoted context omitted.

Its probably that in 1 or 2 years local (free) models will completely take the place of cheap models so cheap models need to move up the quality chain. You have free local models for most tasks, $20 subscriptions for near-frontier intelligence, and API per token costs for frontier intelligence. Flash seems to be targeting the near-frontier category.

That might work if it wasn't for FOMO. Are you ok with only $20 of frontier usage a month?

Subjective, but if we compare to compute not everyone needs the most expensive laptops or super computers for their work.

I think frontier models will be invaluable for scientific research, defense, financial analysis and such. But the average person probably would be reasonably well-served with a local model.

If you're in sales, customer service, product management and such - the leading open models at the 30B mark are already good enough.

Re: Gemini 3.5 Flash

#448

Post by hedora »

eis wrote: post #40

3.5 Flash was more expensive than 3.1 Pro to run the Artifical Analysis test suite. $1551 for 3.5 Flash [0] vs $892 for 3.1 Pro [1]. That's 74% more cost while ranking lower. It's 2.5x as fast but I don't think the bang for the buck is there anymore like it was with 3.0 Flash. I'm a bit bummed out to be honest. I did not expect such a huge (3x) price increase from 3.0 Flash and I bet many people will not just blindly…

Ouch. That's going in completely the wrong direction.

How many people complain that we have too much low quality AI output for humans to read, let alone evaluate vs. how many people are complaining that they want higher quality, more trustworthy output?

Re: Gemini 3.5 Flash

#449

Post by emp17344 »

sosborn wrote: post #437
irthomasthomas wrote: post #161

Earlier quoted context omitted.

This is a perfect illustration of something I noticed with llm progress. Ask them to improve an svg like this, and it never fixes the missing crossbar or disconnected limbs, it just adds more stuff. In this example they have obviously improved greatly, and it contains a ridiculous amount of detail, but they still to get the basic shape of the frame wrong. It's weird. And the pattern shows up everywhere, try it with a…

This matches my experience with human too FWIW.

Why is there always an identical reply like this when anyone criticizes LLMs?

Re: Gemini 3.5 Flash

#450

Post by hedora »

charcircuit wrote: post #402
lugu wrote: post #149

Earlier quoted context omitted.

What we need today is a standard local API (think of it as a POSIX extension). So that each desktop app that needs AI to enhance a feature can simply call that. This way, those apps will need to handle the case where AI is not availabile. This will empower users.

All major operating systems Windows, macOS, iOS, and Android have local APIs for using AI.

Why would I use those instead of just grabbing a model from hugging face? Are they as good as qwen 30B?
Post reply on HN