mosselman

I am always afraid to share what I am actually reading right now because I am afraid that people spoil it for me. I am highly sensitive to spoilers, so even things like "I loved that it was so funny" or "The ending was great" or something just ruins the experience, at least somewhat.

So I'd usually share what I read before this.

Which in turn is essentially the Arc browser's UI

But it doesn’t. That is why that other person is claiming their demo is more impressive.

What you are saying is like me losing 13kg is as impressive as having the potential to do so if only I would stop eating ice cream and too many snacks every day.

That is the totally irony.

I was trying to get fable to analyse the security of my own app to make it safer, but then it started refusing me because of safety rules.

So it CAN help me writing the code that needs to be checked in the first place, but it can’t help me clean it up and make it safer.

I have not worked with Go, so I thought I'd ask: why wouldn't simple find and replace be able to fix this? And why wouldn't a coding agent be able to do this for you quickly?

Exactly. The only metric you need to look at is the cost per task and take your pick from the models that match your budget there.

Spark is such a bad model, you are far better off using Astra Medium than using Spark.

You shouldn't ASSume a link with a name like this is sfw.

What I don't get is, why would we still use Fable now? What is its reason for existing? If it is more intelligent and cheaper that is. Why are they advertising it as the model to use for when you really have to think when their benchmarks show Opus 5.5 is better at everything?

Shameless plug for a server monitoring macOS app I built. I built that too for the servers I managed myself. Instead of working through a service or a self-hosted monitoring tool that I then over to monitor itself again, I thought running it on my laptop was pretty nice too. One of the next features I've been wanting to build is uptime monitoring.

https://kitaso.app/

Sure there is the downside of having to be at your laptop or having it on, but the upside is that it has very few moving parts and is very simple and it just has one set price.

Why is this website asking me to enable notifications?!?

They've started hosting GLM-5.2, that should be very telling of their capabilities at the moment. Hopefully the investment will allow them to hire the right people to become competitive.

"You may also like $250 AirPods"

I posted more on the other branch of this thread. But cost prohibitive is about all I can find about this and not illegal. Which for some people might be the same result of course.

From what I can find online that doesn’t seem true. There are import restrictions and various other restrictions, but they don’t seem to be illegal.

Then again I don’t live in Texas let alone the US so i might not know where to look and I don’t care enough to truly find out.

I was just surprised that they were supposedly illegal. Which seems untrue.

Why are they illegal? What is your source?

Homeowners can’t put solar panels on their roof to use the produced electricity?

Cool. I went another way for monitoring my servers and built a local macOS app instead of installing the monitoring dashboard on the server itself.

https://kitaso.app/

It monitors any server you add to it and it can install a daemon that collects metrics for when you are offline. I keep it running to monitor our servers.

I setup https://grimmory.org/ a week ago. It has a feature where it can sync with kobo by just editing a single config file on the kobo itself and it works great!

I've been building my own metasearch engine after using Kagi for a long while and I must say that my results are getting better even than theirs.

Kagi can do things at their scale that are not feasible for me right now and one of them is categorising links like they probably do for this feature. So it is inspiring. A route that might work is checking if there are ublock lists that keep track of paywalled sites.

I've also been looking into creating my own search index, but this is a huge undertaking. I might go the tiny index route like Marginalia has done and try to find a niche for my specific use cases.

Another route could be to categorise websites better automatically and build up a category index.

Let's assume the calculation is correct and you actually earn back in 11-13 years. You are better off investing the money in something like the S&P 500 where you'd double your investment in about a decade without the hassle.

Things will start to get interesting where €1600 don't buy you 5kWh but double or triple the amount.

I can imagine most people are better off either investing like I said or in insulation.

When I tell Claude or open code using Kimi or codex to fix something on my raspberry pi, they all go to ssh right away.

I found that most things that are “features” in harnesses and tools that don’t relate to UI and UX in those harnesses themselves can be done through skills or just prompts.

A nice features of Claude’s is the remote control through the mobile app. That isn’t just a skill.

You are describing 'the blame should be on Windows'.

The consequence of Windows having the blame is that one should not buy it.

This is so cryptic that I wouldn't even know if for or against would mean for or against 'chat control'

This is an incredibly good idea. 90 days is too short to feel the effects of some things though. Better make it a year.

The downside is "lets try giving everyone basic income of $100k/week". But apart from that great!

I think Anthropic has been really successful in their branding of Mythos as something, mythical.

I never heard anyone contemplating "why people are so sure we'd get freely available GPT-4 level+ models".

Mythos is just another advancement in AI model capability. If we start pretending like it isn't, we are setting ourselves up for paying far more than is reasonable and accepting the frame that we should not in fact have free access to open models with the same capabilities.

This is just marketing.

I can't wait for some similar big steps in open weight models.

Update since the side thread blew up.

There is some difference in how OpenAI and Anthropic handle 'max_tokens'. The OpenAI way raises errors in Anthropic for example.

I will look at confirming this a bit more in depth once we've taken our RubyLLM based adapter in production and see if I can make some contribution.

Thanks for all the work! It is incredibly impressive and I never meant any disrespect.

Thanks for the discourse. I never meant it to be disparagement nor do I think it really was.

I checked and it turns out I remembered correctly that setting effort and some of its settings are not portable between providers.

There are some different settings that each provider uses and in order for it to be portable, you have to force some defaults on provider A when using a setting that is almost only supported in provider B.

In our implementation we decided to drop a certain setting when using OpenAI in one case and we decided we can just force some other setting when using Anthropic. But this 'solution', might not be what others expect.

When you build an open-source library you can go this opinionated route and force these settings, or you might go the config route and force people to explicitly handle per-provider differences. I will have a look at what I am able to do in terms of a contribution and then in the PR Carmine can decide what they like.

We had already implemented tool_calls in our own database and have a system that executes them and creates our conversation array, etc. So we wanted to leverage the providers that RubyLLM supports without having to change the tool execution in our platform.

Hi! Valid challenge, I am probably misremembering. We were playing with various 'one-interface to all providers' solutions and I might have mixed up RubyLLM there. Sorry for that.

I will have a deep dive into which things I felt we needed to adapt per provider.

I didn't mean to imply that you have to solve all of our wants of course.

One thing we did do was monkey-patch the spot where tool_calls are performed by RubyLLM. We had our own mechanism for that and were able to skip RubyLLM's and still extract the tool calls and run them through our own tool harness. That all worked beautifully. I don't know if that type of stuff is something you want PRs on or that you want to keep steering towards the route that does everything within RubyLLM classes. Happy to contribute some of that.

It is quite nice, but not as nice as you'd want. You still have to set platform specifics when running completions when you want to tune things like temperature, effort, max tokens, etc.

what a nonsense, generated, article.

> For context: GLM 5.1 ran the same task and reached 7.3x. Kimi K2.6 reached 5x. DeepSeek V4 Pro reached 3.3x. The models that stopped early did so because they issued no tool calls for five consecutive rounds, they concluded they couldn’t make further progress and stopped. Qwen3.7-Max didn’t stop.

By this reasoning I could release a model that lacks all the basic optimisations. Have it optimise itself for hours to reach 20x the throughput and then claim that the model is superior to the others?

I am not saying that is what happened here, but the reporting is abysmal.