Join the discussion
Write your take first — we'll ask for email only when you're ready to publish.
- Hacker News
- Why does an open weights model cost nearly the same as GPT5.6? $1.14 vs $1.23 on the cost index. Since you can't presumably run this on your own hardware given the model size and hence gain other things like privacy, I don't see any reason to move away from GPT at this rate.by drnick1
- You said it yourself, model size and hardware. Big models cost more (good optimisation reduces things slightly, but they still need the hardware).by Alpha3031
- Why should open weights correlate with cost? Cost correlates with the expense of running the model more than it does to the expense of developing the model.by ecocentrik
- Why does an open weights model cost nearly the same as GPT5.6? $1.14 vs $1.23 on the cost index.
What cost the most in API. Input, Cached Input, or Output. There you have your answer.
Unfortunately, we have moved so much of the actual intelligence of models towards reasoning, what results in some models getting good scores, but this is because they are dumping a insane amount of reasoning tokens at the problem.
So a mid priced model, with heavy reasoning output, cost the same as a expensive model, with medium reasoning output.
Before the GPT Luna price drop of 80%, you actually had the same price if you used Luna High and Sol Low. With the difference that Sol Low was insane fast, and often way better code.
Do not look at the top score but more what is on the horizontal axis as you go down. Sol Medium is frankly, was the best performance for dollar, until that Luna price drop. I will even argue that despite the higher price, Sol Medium is still way better despite Luna Max being cheaper. Or Opus Low, one of the better values also.
What do you notice? Is that those models all have a high intelligence start point for their low setting. So that means they do not rely as much on output tokens aka thinking.
by benjiro29 - Qwen Max is their large model - over a trillion params. Similar to Kimi K3 in size. Qwen 3.8 27B is going to be more accessible to your own hardware. I'd say that Qwen Max is not approachable for the majority of people and companies to self-host.by jjice
- For one thing, providers of open models can't arbitrarily increase their prices without facing competition.by apitman
- It's not enough that it's better?
Many providers will host it and will compete on price. It also can't easily be taken away because one company (or one government) decides they don't want it around any more. People can fine-tune it for particular workloads.
by eli - GPT5.6Sol completes the suite in 70M tokens, while Qwen3.8Max needs like 145M tokens. So this is a case where models like Qwen 3.8 and Kimi K3 use a lot more output (reasoning) tokens, go a good bit slower, so they can ultimately achieve a better intelligence score than if they went more quickly.
There are a couple of frontiers (ok bad word, maybe categories) in open weight models.
These Qwen 3.8 and Kimi K3 style models aren't trying to win on price, they're trying to compete on intelligence and capability.
Models like Deepseek V4 Flash (updated this week) are $0.03 a task, or 50X cheaper than Qwen3.8/Kimi K3, and 100X cheaper than Fable, while offering stunning intelligence. That's a different frontier for competition, and perhaps one more interesting for someone who wants to see them compete on cost.
by criley2 - A couple days ago they had published an overall score of 53 for this model, but that was removed and today it returned with a score of 56.
I wasn't able to find an explanation from them. Anyone knows what happened?
by quirino - A wire transfer happened.by Art9681
- according to them:
> launch traffic hit our public API endpoint harder than expected, causing intermittent instability.
by whwhyb - I find that surprising.
I've been trying it on several projects and have found it's pretty sloppy. It leaves stuff broken, doesn't reliably write tests to check its own work unless explicitly prompted, misunderstands the assignment, etc.
It is smart and reasonably quick but not reliable.
by SwellJoe - It’s because they’re doing some sort of combined score of intelligence, speed and cost. On pure intelligence it doesn’t even show up in the top 10.by dyauspitr
- I've come to the same conclusion over and over with all of the Chinese models that have been claimed to be catching up with OpenAI's and Anthropic's frontier models (Deepseek 4, GLM 5.2, Kimi K3).
At their best, I think they're closing in on Opus and GPT, but they're incredibly inconsistent and the variance in output quality is much higher than the best from any of the Anthropic or OpenAI models from the last few generations. The only way I can describe it is that it feels like a lack of intuition with the models which means I find my self needing to write longer prompts or have more back and forth to get them to do what I want from them.
To give an example, I have a saved prompt that I use as a sanity check on some data I'm storing. It reads about 50 rows from a DB and matches them to the UI and makes sure the data is displaying correctly. I've been using this with GPT 5.5 and now 5.6 for a few months and running it a few times a week with no issue. Sometimes I'll run it multiple times in a single chat if I notice bad data (run it, fix thing, run again, fix another thing).
I recently tried to switch to using Deepseek v4 (first flash and then pro) and while both did the task just fine, both would do things like change the response format from one message to another in the same chat or randomly decide to omit things it didn't think were relevant. At one point I ran the prompt, fixed some bad data, and then said "Okay, I fixed row 7, run {prompt} again" and so it decided to leave row 7 out of the response. A few times the first message would contain a table and then the next run in the same chat would contain the data in a bulleted list.
None of those are major issues and all could be solved with a bit more rigor in my prompting, but for me it makes them harder to work with. Those examples are a bit trivial, I think they're the easiest way for me to illustrate the gaps I see with them.
by superfrank - I am so excited for Qwen 3.8 27B. It’s a shame how slow prefill (~3-400) is on a strix halo but it’s such a good model for agentic tasks.by syntaxing
- What type of agentic tasks are you using it for (eg how complex)?by tarr11
- How are you running it on a Strix Halo? The weights aren't out yet, are they?by CamperBob2
- I find that 35B-A3B is much easier to run on my M4 Max (both prefill and generation)by LoganDark
- Prefill is survivable if you cache well. But what kills me is the context. Qwen 27 needs a ton of room for KV Cache. I guess not an issue on a 128 GB Halo or Spark, but if you are running of consumer/prosumer GPUs it's miserable to be compacting every 120k tokens.by colingauvin
- Hopefully this boils down to the smaller versions they've teased. In my experience, Qwen models are the closest to the "less knowledge, more intelligence" (yes, the two are hugely correlated!) ideal some tool-dependent tasks need. Even the 3.5 2B can be easily prompted to always lean on tools and not jump to false conclusions (although its actual coding skills are abysmal, as you'd expect).by petercooper
- > less knowledge, more intelligence
People produce such models by over-RL-ing smaller models on math and coding tasks. I've found the results capable of neither innovative work nor thinking outside the box. They're straight-A students raised by tiger moments who never let them play freely for hours in the dirt.
Perhaps you could say such models are skilled --- but intelligent? Not by my measure.
People and AIs alike need diversity of experience and a broad liberal arts education to see hidden connections between fields and make real advances.
by quotemstr - Anthropic is a bit nuts, I had $260 of credits on my max account for the extra usage the other night. It was expiring, so I figured I'll fire up an agentic swarm to deep dive and make some deep changes to some old cold bases.. literally 25 minutes or less, $260 burnt, it didn't get get into the implementation, just wrote a ton of useless plans for the most part. It really opened my eyes to what they expect to charge people.. wayyyy overpriced.by theropost
- You plugged in a space heater on a roofless house.
There is some element of responsibility on the user to guide and monitor the model/harness and not let it rip to burn tokens.
by hahahaa - I had same experience with OpenAI. I have the $200/month plan and use 5.6 Sol all the time. What would normally use about 2% of my weekly allowance burned through $100 of credits in 40 minutes.by brynnbee
- Anthropic is the new AWS.
Amazon's first principle is the Customer Obsession. Making customers happy.
Fun bit is that the human psychology rates personal looking fixes better than having no issues at all.
For example, AWS overcharges you, you contact support, and more or less hassle free they refund or issue credits. The customer feels appreciated, or at least got something "extra" or "special treatment".
Meanwhile, any other (small) cloud. Simple, no weird charges. Even _most_ of network egress is free. But, no reason to call support or feel "extraordinary". Comes out as "meh" against Amazon's "top tier" support model...
by pvtmert - It’s crazy how different the credit cost and subscription cost are.
With the $200 subscription, I can have Fable on ultracode working for hours and not dent the usage limits.
by cortesoft - And at that cost they're still not profitable. It's going to be a bumpy road ahead...by mikae1
- Its tough to go from max account at home and pay per usage enterprise account at work with heavy usage limits... but the limits are there because pricing is insane. Feel like I'm in the $5 Uber rides phase at home.by swalsh
- I managed to lose around $300 in credits I had saved for some emergency /fast sessions the following way: switch to Fable. Work on the design. Downgrade to Opus for the build. If any of other parallel Opus session has /fast enabled it seems to enable it for the newly spawned session by default. Before I knew it, the $300 was gone. I think the bug is now solved, but it was rather unpleasant. I dont ever remember bugs that would drain my wallet - with claude code its just another Tuesday. Still love it.by aenis
- Hint: The new models are really good at burning tokens.
I've had to use it a bit for work, and it's been remarkable watching the degradation in performance with the default suggested current models (Opus 5 as a prime example) vs the models that got them huge attention a year ago (Opus 4.6)
If you give 4.6 a spec, or existing code to implement a feature in, it will ask some pointed questions if there's something unclear in the spec, and then produce a plan and move to implement it.
5 will freak out at even a basic task, ask itself if it's own assumptions or your instructions are correct, proceed to re-assess it's own plan, and it's instructions 3-4 times, and then maybe produce code after burning several hundred thousand tokens (and quite a bit of time) analyzing existing code and thoroughly sweeping it for irrelevant problems both to the task it was given and the spec it came up with.
It's quite bizarre to me how well advertised the benchmarks and anecdotes from people one shotting MVP browser games are, compared to the experience of everyone I know that's had to actually use it to accomplish even a relatively basic task.
by tarnith