Hacker Newsnew | past | comments | ask | show | jobs | submit | nsbk's commentslogin

This so much! What great games, probably my favorites over the early FPSs of the time, but in close contention with Day of the Tentacle and Monkey Island

Spain is at 96.79% (June 2025), up from 94.79% a year before. See [0]

I think France and Spain are the best of the big European countries regarding coverage, but also adoption: If we look at the adoption ranking globally, Spain sits fourth at 90%, behind the UAE, China and South Korea and ahead of France at 83.6%.

Very few European countries convert coverage into subscriptions that efficiently, and I think it's driven by competition. You can get 500Mb symmetric connection for as little 10 euro [1], 20 for 2Gbps. DIGI has just gone crazy and in their own network they are offering symmetric 10Gbps for 25. Most people I know have 1Gbps because when is that not enough?

[0]: https://digital.gob.es/en/telecomunicaciones-infraestructura...

[1]: https://www.digimobil.es/fibra-optica/


In Italy we are at 78% of addresses covered. That doesn't mean automatically that one gets fiber. First at all, if someone doesn't need to download heavy stuff and has a good coverage a 5G SIM with unlimited data (something you can get for 10/15 euros a month) is cheaper than a fiber connection (that starts from 25 euros).

Second a lot of people has still FTTC even if it has FTTH, mostly because FTTC offers a speed that is enough for most people (100Mbit/s) and upgrading may have costs (the installation of the fiber is at the expense of the ISP, but if you need to make modifications, such as put a pipe from your house to the street in which they will put the fiber, it's at your expense). In some cases even if fiber is present it's difficult/not possible to get it into your house (for example you live in a condo and the other tenants doesn't want to spend money to make the modifications that are needed to pass the wires).

In any case they have started to switch off copper (since it costs money to maintain an infrastructure that has decades), so either way customers will be forced to move to fiber or 5G network in the upcoming years. In any case a fault on a copper line is likely not to be repaired...


> a 5G SIM with unlimited data (something you can get for 10/15 euros a month) is cheaper than a fiber connection (that starts from 25 euros).

These sorts of prices are unheard of in the US, so incredibly low.

Something has gone deeply wrong with US telecommunications.


You can use any one of many MVNOs to get unlimited data for around that price, say less adjusted for household income, in the USA.

A favorite of mine is https://www.usmobile.com/ which is $25/mo list price (and less if you pay for the year, or if they're having a sale) which will give you Unlimited Data and 1GB a month of global roaming data

If you pay $45 you can get unlimited data from Mint for $15/mo: https://www.mintmobile.com/plans/

It's not unheard of!

It's actually the high speed internet at home that is less competitive - the prices in Europe are unheard of.


Only some places. Netherlands is €45-60 for 1 gbps (we don't make 4x-6x the money Spaniards do), unless you live in an area that actually sees multi-line competition.

We do have an interesting sidegrade: 5G Klik & Klaar from Odido (formerly T-Mobile / Deutsche Telekom) for €26. If you point it the right way out your window, you can get 700-1000 mbps down and 150 (capped) mbps up. Ping is around 35ms. Horrid modem firmware though, you get ping spikes up to 400-850ms (!!) under load due to bufferbloat, but you can of course solve that with your own router and SQM :+)


Is this line of site to a very high mast? Enabled because the Netherlands is so flat?

In the U.S. I perceive MVNOs to be a scam. Like Tracfone was OK in 2011 but somewhere afterwards it became normal to go around rural upstate NY and no service and go around NYC and no service and go to LA and no service.

This year I got a real phone with a real plan and I have coverage in many places now me I can finally see why people get excited about apps.


That's kind of a broad brush, Google Fi is great, especially the fact that they work mostly seamlessly for the same price in like every country, and they're an MVNO.

It proves that you can’t trust the claim that “it is the same network”, I mean it is the same network but you may find yourself riding the back of the bus and there is no independent way to check what service you are actually getting.

Well I never expected it to be treated exactly the same as their first party customers, did you? Especially if I’m paying a much lower rate. Just have to evaluate the actual results, and hop if you’re unhappy. I’ve been extremely happy with mine, so I’ve stuck with it. It’s a bad choice if you’re out and about a lot, since it’s not cheap when you go unlimited data, but working from home on WiFi, the metered option, I end up paying like $30-$40/mo, and there’s no need to go get a new sim when I travel. For a different lifestyle shape, I’d be shopping for one that fit that better.

I have both T-Mobile via a family plan and US Mobile (AT&T MVNO) via a yearly plan.

US Mobile serves as my data eSIM, where it gives me unlimited hotspot 20GB of roaming/month. Previously I bought eSIMs when traveling since T-Mobile is only 5GB/mo hotspot/roaming and earlier at 256 Kbps.

T-Mobile is better for calls and international texts, due to it being unlimited. I also prefer keeping a non-MVNO as my call/text number so I can go to a T-Mobile store and activate a spare phone if I lose mine (which I never did), or my phone fails. With US Mobile, I'm not sure how I'd recover a lost phone.

Since the Pixel 10 Pro lacks a SIM card slot, this is important. But hey, I used my Pixel 3 on Sprint with eSIM; I went T-Mobile 3 months before they merged.

US Mobile has done well with me, although anything not-T-Mobile is really slow nowadays.


I do use US Mobile, and I recently switched to it because T-mobile made it so difficult to pay a bill, it was nearly impossible to get them to accept money.

And a ~decade ago in SF I did have Sonic 1Gbps symmetric fiber for something like $40 or $50/month, but that too is twice the cost of Italy.


Can't speak for Italy, but there have been some consequences in many EU jurisdictions. The EU has been pushing HARD for competition, etc, but it has left many European telecoms in debt-heavy financial situations that has caused issues around future investment and in some countries the (mobile) coverage is good on paper, but the networks are themselves actually terrible (looking at you Germany).

But yeah, it's otherwise a problem with competition enforcement.


At the same time a software developer earns 40-50k, perhaps a little higher in Milan or Rome.

State sanctioned and encouraged anti trust behavior will do that

> These sorts of prices are unheard of in the US, so incredibly low.

The EU also has incredibly low salaries.


Gone? It started rotten with Bell and hasn't gotten much better.

Because deploying fiber is so hard in NYC, I use Calyx Institute (https://calyxinstitute.org/membership) on a MikroTik Chateau 5G R16. Calyx uses T-Mobile towers. I use the internet via L2TP to a VPS to give me a static IPv4 and IPv6 /56 block. This is only for upload speeds. I get 45-80 Mbps uploads.

I used Spectrum for four months but because of my L2TP setup, DOCSIS ACK suppression doesn't work; it's TCP inside L2TP UDP and then ACKs are then sent over the wire. This meant sluggish internet under load. Spectrum did install a high-split amp but it's still not turned on afaict.

Row townhouses in Brooklyn and Queens have Fios, but not in Manhattan. I wish I could just do a NYT ad but that'd force me to liquidate a decent chunk of my E*Trade and J.P. Morgan accounts.

I am right now visiting Belgium and I saw multiple towns, including historic neighborhoods with Proximus fiber splitters: Brussels, Antwerpe, Ghent and Bruges. The fiber and Telenet Coax is so well organized when compared to NYC's coax mess. However, where Fios is available, Verizon does an okay job with organizing fiber.

Verizon said killing line sharing would bring more fiber. Yet Proximus built fiber in historic cities more than Verizon did. I don't believe Orange France legally has to share fiber, but in return Orange can't block Free/SFR with payola deals.


> In Italy we are at 78% of addresses covered

> UK is at 85% FTTH [on a sibling comment]

Meanwhile I have 1 Gb/s FTTH in an Italian village of less than 100 souls, while I couldn’t reach 300 Mbit/s on copper in London zone 4.

I have been surprised at how well they have improved the situation in Italy, and in dismay at how terrible it is in UK.


It’s not uncommon for cities to be worse than rural areas now. In Ireland, nearly all rural premises have 10gbit-capable fibre available (will be literally all by early next year) but parts of central Dublin are stuck with VDSL. Retrofitting in cities can be quite slow and expensive.

Average speeds are higher in the UK than Italy.

My brother in the UK has 1Gb/s fibre in a village of less than 100.

https://en.wikipedia.org/wiki/List_of_countries_by_Internet_...


Well I guess I must have been unlucky in 13 years in 5 flats around London with an average broadband speed of ~150 Mb/s

Clearly, yes. Unless you're claiming your personal experience invalidates a long running, well regarded statistical analysis.

No, you're totally right. I'm currently in Zone 4, maximum _theoretical_ speed is 67 Mbps.

The UK is absolutely terrible. I've noticed this quite a bit in my travels there.

The main issue with copper comes from the active infrastructure rather than the physical lines.

I work at an ISP that gets DSLAM equipment from Huawei, we run one of the few G.fast VDSL deployments in the world using this particular hardware. Huawei has stopped manufacturing this product line (the Swiss market is a rounding error ;), which will force us to fully migrate some neighborhoods, only for backup inventory.

Germany being 15 years late, I assume you can still buy similar hardware from AVM though.


I am in Italy right now, in fact, spent a lot of time here. I can safely say, the 5G sim coverage is extremely spotty, and when you can get internet from a physical line, it's generally pretty poor speeds. I am pretty sure a lot of that 78% includes 1 bar of LTE coverage, so it's "technically" true.

Several of the issues you point have been solved in France via regulations (on the ISP and on the tenants of condos). It tools several years, even decades, but we ended up with fiber regulations that are good (could have been better and faster).

> In Italy we are at 78% of addresses covered

Complete bullshit.

BUL maybe is close to that percentage

P1G is nowhere near and is never going to reach full coverage in 2030.


I spent time in the interior of Portugal that had fibre internet, it was very impressive and incredibly helpful to the village. Better payment services and banking for example.

UK is at 85% FTTH or 91% gigabit (including DOCSIS): https://labs.thinkbroadband.com/local/. Given VM are a fair way through their XGS-PON upgrade it'll probably be 90%+ FTTH imminently.

Damn, we pay $110 for 5Gbps in Seattle only because the 2Gbps would be $100.

Cellular plans are also a lot cheaper in Europe :(


I pay £6 per month for a sim only contract, which gives me 4GB of data. In the USA I feel like I was paying closer to $40-50 for a google fi sim? Which admittedly did have better roaming. Now I just get an e-sim for travel and it works out to be £3ish per day

I’m in the same camp as you. I’ve just decided that the M5 Ultra is gonna be my next machine. Plus an Air for the few times when I need a computer on the go. I’m invested in Tailscale already so an ubiquitous Studio Ultra seems like a sweet idea

I don’t even know what a Mac Ultra looks like and I want one.

It will be new Mac with more GPU cores for ai, right?

It would sit in my basement server room providing local inference.


I checked eBay and they are now around 200 usd and up. All the sellers located in the US and Canada, so a bit more involved for hackers looking into building one from Europe. I still pulled the trigger on one, not sure I will use it much but I'm definitely gonna have a lot of fun building it!


This is what I have been waiting for in order to process my massive action camera library, I will definitely give it a try!

I will start with older standard GoPro videos, and hope that by the time I make it to the Insta360 videos the editor supports 360 video (unless it does already!)


not yet with .360 files. we currently support .mov and .mp4 for videos!


It is 17% more token-efficient than 3.5 and performs significantly better in coding and tool usage benchmarks.

It is also cheaper than 3.5:

> This enhanced efficiency is also combined with a lower price than 3.5 Flash. At $1.50/1M input tokens and $7.50/1M output tokens, 3.6 Flash reduces the overall cost per agentic task, making agents more cost-effective to build and run.


This feels a lot like free spins in slot machines. I have a friend obsessed with using up all the Fable usage which supposedly ends today. He's doing a lot of stuff of questionable usefulness just because it's a resource perceived as scarce.


Yeah, it's the gacha of software development. I too redline my subscription every single week just because I want to get the most value out of the money I spent. Literally 100% weekly usage, every single week.

I am using the resulting software and documents though. I'm daily driving the projects I wrote with AI. Started new ones for fun, such as decompilation projects for my cherished childhood games. Hardened my router and my LAN. I just keep finding an unending amount of things to do.

It took over a month to complete a full Fable code review of my lone lisp project.


That's what half of people's AI developed projects feels like to me. I have colleagues using Openclaw, and having an entire team of agents managing their home infrastructure. And I'm just thinking "why?". They're not developing anything that useful, their homes are already automated via home assistant. It's just agents for agents sake. Outside of my work life, I find very little need to be using AI.


Remember when people used to play Pokemon Go?


AI in general is a slot machine and it wouldn't surprise me if that was a large part of the psychosis. You prompt the lever, watch the display spin, and sometimes, just sometimes, a bunch of coins come out.


There's no psychosis in noting that something consistently produces good work, or in wanting to get the most out what is effectively an employee - hired for the price of the subscription. Seriously, if that's 'psychosis', then every employer on the planet eager to wring out productivity from their employees is afflicted with the same psychosis.

Isn't it rather a psychosis to insist that people are imagining the results they see, just because you don't like the tech and can't stomach the thought there might be something in it?


We’re at the gamification stage of motivating people to use your product, eh?


Permafable got announced earlier this week, but definitely the slot machine feel is real.


What do you mean by Permafable?


"Beginning July 20, Claude Fable 5 will be included in all Max and Team Premium plans, at 50% of limits.

Pro and Team Standard users will continue to have access to Fable via usage credits, and will receive a one-time $100 credit.

Demand for Fable has been challenging to predict, which is why we rolled it out to subscription plans in stages, extending access several times as we secured additional capacity." [0]

[0] https://x.com/claudeai/status/2078302415804379218


I am assuming he means that Fable will remain part of the subscription plans going forward, but only for Max and Team Premium plans. For other plans it will be available via credits only.


Why does Fable usage end today?!

Anthropic won't have anything competitive with OpenAI Sol and Kimi K3?


Bring it on! Hoping that they release smaller sizes of Qwen3.8. I use the 35B MoE and 27B dense models locally and most of the time I don’t need to reach out to Claude. Extremely useful specially when requests include sensitive and/or personal data


I think everyone is hoping this!

It would be great if they'd release an MoE model somewhere between the 35B size of 3.6 and the 122B version of 3.5 - it could be a great balance of speed and ability for people with reasonably powerful but not insane home computers.


> the 122B version of 3.5

Yeah, this is what I'm holding out for, the NVFP4 variant of 3.5 122B is blazing fast with reasonable quality and even with max context fits perfectly within 96GB.


Sweet. Are you running a Mac Studio Ultra or 4x3090?


Simpler (and faster): a single RTX Pro 6000 :)


Curious, do you find the 3.5 120B sized MoE works better than the dense 3.6 27B?


Yeah, Qwen3.5-122B-A10B-NVFP4 produces better responses than Qwen3.6-27B-NVFP4 (both from unsloth), but I'm mostly using them for programming in various ways, mostly Rust, Clojure, Python and JavaScript, and some translations tasks, but not much more than that, so YMMV.

Edit: as a concrete example, I'm working on a "optimization framework via agent harness" right now, Qwen3.6-27B-NVFP4 is often unable to actually complete the optimization within 100 turns, while Qwen3.5-122B-A10B-NVFP4 has no issues finishing within ~50 turns or so.


> Yeah, Qwen3.5-122B-A10B-NVFP4 produces better responses than Qwen3.6-27B-NVFP4 (both from unsloth), but I'm mostly using them for programming in various ways, mostly Rust, Clojure, Python and JavaScript, and some translations tasks, but not much more than that, so YMMV. > > Edit: as a concrete example, I'm working on a "optimization framework via agent harness" right now, Qwen3.6-27B-NVFP4 is often unable to actually complete the optimization within 100 turns, while Qwen3.5-122B-A10B-NVFP4 has no issues finishing within ~50 turns or so.

But why are you using Qwen3.6-27B-NVFP4 compared to the FP8 or full version? In my experience the Q8 of 27B is on par sometimes better than 122B. I am experiemnting witb higher quants for 122B to fit on my Strix Halo, but still, the difference honestly for my workflow is not that much. I just wish they released 3.6-122B version.


dense small models do not like quantization. i find 27b fp8 to be smarter albeit less knowledgable versus the 122B


Hahaha well played! Hat tip


Absolutely. There's a glaring gap in the space for something about the size of Nemotron Super or just under, but actually ... competent.

The fantasy is a 100B or 80B model, but MoE and highly tuned for coding.


SO MUCH THIS. I have Strix Halo with 128GB RAM and was a large and fast model like 122B A10B. Here's hoping!


Indeed! That would be the sweet spot for my 2x3090 rig


This seems more of a battle for frontier AI supremacy. I'm afraid that small capable models have been left in the dust. Big labs don't really want to hand over the golden eggs goose to the end user. Possibly the hardware vendors(e.g. Nvidia) may want to play in that area as well, to pull money from all parties.


> I'm afraid that small capable models have been left in the dust

I wholly disagree. Rather than going the "everything is a claude code skill" route, I've been hacking together purpose-built harnesses for all sorts of tasks, and in that environment a wee little baby model can do some really useful things. You end up burning lots of tokens making the thing, but then all that investment comes back when the resulting tool works perfectly fine on a dinky little model that fits on my 3060 Ti.


Google makes Gemma 4 31B QAT; that's not a small lab. It's one of the better models out there for consumer hardware. Allows me to run it on a 7900XTX with 64k context.


someone will keep putting out consumer level models. Once you have the larger models, you can derive the smaller onces.

Europe will definitely be interested in democratizing these things if China starts losing interests; from there, there'll be more countries looking to keep their citizens entrained in their own Country's infrastructure.

It'll especially be true if the memory cartel keeps prices high and NVIDIA tries to gouge higher memory models.

It's an arms race everyone can join because PC hardware was mostly democratized in the last decade.


nvidia and amd don't give a flying fuck about end users right now while they can milk triple digit markups from infinite money VCs via data center GPUs.


That's a 2.4T model, how would they reduce this to 35B and still give some accuracy? That's a completely different arch.


there's been a lot of research about reducing models by taking out layers; there's also using it to train smaller models by optimizing parameters.

I dont see most model building as anything more than a pig at a slop troth, despite the level of sophistication; they're still rarely pruning the input beyond random sampling.


What hardware do you have?


I run a 2x 3090 rig, but a single 3090 already provides a great experience at a reasonable quant and context size. On a single card I used to run Qwen_Qwen3.6-27B-Q4_K_M or similarly quantized 35B MoE at 65536 context size



Such a good read. Makes you think. Thanks for sharing!


My brother and I restored my dad’s ‘64 Vespa 150S and the thing is extremely loud!

It is very smelly as well as it’s a two stroke engine, but I don’t mind that at all. Quite the opposite


Agree, as long as you don't use it around other people it's probably fine.


A friend races Vespas. They are apparently well suited to tuning (perhaps being poorly tuned from factory?), and power is much increased. The noise seems to go up vastly.

He has to re-weld the frame from time to time due to the cracks that form.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: