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

DeepSeek V4 – almost on the frontier

simonwillison.net

Post reply on HN
Print view View original

151–160 of 420 posts

Re: DeepSeek V4 – almost on the frontier

#151

Post by curioussquirrel »

V4 is definitely a step-up from V3.2 on our multilingual benchmarks.

Two caveats: - when inferring through Openrouter, we've had a lot of issues with very slow speeds (TPS) and an occasional instability. I just checked and it's still 10-30 TPS on all available providers, which is not a lot for a model that likes to think as much as DeepSeek does.

- the official DeepSeek API makes no guarantees of data privacy even for paying users.

Both points could be moot with using it through Azure AI foundry (the latter is, afaik); I have yet to test that.

In any case, happy to see more open-weights models that are somewhat competitive with SOTA models!

Re: DeepSeek V4 – almost on the frontier

#153

Post by segmondy »

piker wrote: post #129

Jensen has a point. I believe these were trained and run on Huawei chips. The Nvidia embargo may backfire on American leadership as necessity gives way to invention.

It's too late already, that ship has long sailed. China has the know how in software and hardware. They don't need American tech, they just want it because it's convenient.

Re: DeepSeek V4 – almost on the frontier

#154

Post by gordonhart »

sanex wrote: post #138
cedws wrote: post #113

The biggest differentiator for me: DeepSeek just does what I ask. I've tried using both GPT and Claude for reverse engineering recently, both refused. I even got a warning on my OpenAI account.

We have an enterprise cursor account so I can try all the mainstream models. Using composer 2 on our own code which I obviously have the source code for I couldn't get it to turn on a debug flag to bypass license checks while I was troubleshooting something. Infuriating. It was like that old Patrick from SpongeBob meme. I don't understand why we would turn the models into law enforcement officers. Things that are ill…

> I don't understand why we would turn the models into law enforcement officers

It's a simple corporate risk minimization strategy. Just look at how universally despised Grok is on HN. Not because it's a bad model, but because it has less aggressive alignment which means it can be coaxed into saying things that get Xai pilloried here and elsewhere.

Re: DeepSeek V4 – almost on the frontier

#155

Post by segmondy »

ghm2180 wrote: post #90

I've been using the planning framework from Matt Pocock on very typical brownfield code. I use a harness over claude code, this is so cheap that I would be tempted to mirror my initial prompt to it and compare their responses to the task.

Do you have a link to this?

Re: DeepSeek V4 – almost on the frontier

#156

Post by mannanj »

sanex wrote: post #138
cedws wrote: post #113

The biggest differentiator for me: DeepSeek just does what I ask. I've tried using both GPT and Claude for reverse engineering recently, both refused. I even got a warning on my OpenAI account.

We have an enterprise cursor account so I can try all the mainstream models. Using composer 2 on our own code which I obviously have the source code for I couldn't get it to turn on a debug flag to bypass license checks while I was troubleshooting something. Infuriating. It was like that old Patrick from SpongeBob meme. I don't understand why we would turn the models into law enforcement officers. Things that are ill…

Maybe control is also profitable.

Re: DeepSeek V4 – almost on the frontier

#157

Post by lolpython »

ignoramous wrote: post #150
cedws wrote: post #113

The biggest differentiator for me: DeepSeek just does what I ask. I've tried using both GPT and Claude for reverse engineering recently, both refused. I even got a warning on my OpenAI account.

> even got a warning on my OpenAI account Edit: https://chatgpt.com/cyber

> https://openai.com/cyber

that link 404s

Re: DeepSeek V4 – almost on the frontier

#158

Post by segmondy »

cyber_kinetist wrote: post #17
holysantamaria wrote: post #16

From the pricing page of deepseek: (3) The deepseek-v4-pro model is currently offered at a 75% discount, extended until 2026/05/31 15:59 UTC. Was this taken into account when reviewing the model?

Yeah even the Chinese open models have a problem that inference costs for these aren't that cheap. The only way out for the AI bubble collapse is simply more efficient hardware at lower costs and infrastructure setup downtime.

You can imagine the GPUs cost as fixed, then your costs becomes energy. Efficient hardware and lower costs will pop the bubble faster. The only way out is profit.

Re: DeepSeek V4 – almost on the frontier

#159

Post by Wowfunhappy »

oneseven wrote: post #140
sanex wrote: post #138

Earlier quoted context omitted.

We have an enterprise cursor account so I can try all the mainstream models. Using composer 2 on our own code which I obviously have the source code for I couldn't get it to turn on a debug flag to bypass license checks while I was troubleshooting something. Infuriating. It was like that old Patrick from SpongeBob meme. I don't understand why we would turn the models into law enforcement officers. Things that are ill…

They're probably worried about liability. Let's say that Oracle finds out you reverse engineered their DB using Gemini. You can be sure they will sue Google. Not just for providing the tools, but you could make the argument that it's actually Gemini doing the reverse engineering, and on Google's hardware no less.

Let's say that Oracle finds out you reverse engineered their DB using IDA Pro. Would you expect Oracle to sue Hex Rays?

I don't understand why everything changes as soon as an LLM is involved. An LLM is just software.

Re: DeepSeek V4 – almost on the frontier

#160

Post by pimeys »

100ms wrote: post #108
Havoc wrote: post #97

This gives me hope that when the subsidization circus ends and everyone is on pure usage then it won't be entirely exclusionary to mere mortals who don't have $200pm budgets.

By the time the dust settles I wouldn't be surprised if personal interactive usage couldn't even be had for under $200. I can't fit my modelling of the serving costs of these things to any public reporting, even the more bearish examples

We pay per token in our company. It is not hard to spend $100 for one morning coding session. So thousands per month per programmer. The company finds it valuable enough to pay for, but if I ever paid these from my own pocket I'd look into DeepSeek et.al.
Post reply on HN