alt Hacker NewsBest LLM for every budget, updated daily
Comments
If you have a 24-64GB mac, consider running Qwen3.8 27B locally at night. It's a bit slower to run locally, but if you're sleeping it's less of a problem.
Depending on your memory, you'll need to use the weaker Q4 versions but they still perform well.
It ranks higher than GPT-5.3 Codex (xhigh) or Claude Opus 4.6 (max) so is great for pairing with https://github.com/kunchenguid/gnhf for nightly experimentation, cleanup, or recommendation lists for in the morning.
I'd really like something that's more oriented around subscription fees.
If I want to spend $100 on LLMs next month, what should I do? Get Claude because Opus 5.5/Fable 5.1 are scoring well? Get Grok because 4.7 is supposedly a good mix of competence and cost? Try out a Chinese model? Don't do a subscription at all like this site is saying?
I've been running a quant/tune of Qwen3.8 27B on my M1 Max 32gb MacBook. That plus a good pi setup is having great results. I've used a full q8 of the model before and I dont see a real difference other than how slow it is. But leaving it running overnight on tasks is working great. It is currently debugging some issues in a native Mac Swift application and getting through the list of issues just fine.
This is the one that works good for me on 32gb:
https://huggingface.co/ISTA-DASLab/Qwen3.8-27B-GSQ-RCO-GGUF
Specifically this one: Qwen3.8-27B-GSQ-RCO-IQ3_S-mtp.gguf
Does anyone actually pay API costs out of their own pocket? It's about 10x cheaper to just get a codex or chat gpt subscription, it's so heavily subsidized compared to the API that I'm sure it would be cheaper to use frontier models on a subscription plan rather than paying API prices for deepseek flash.
As the time goes on it only becomes harder to differentiate between model capabilities with just one or two numbers. I would love to see some kind of multi-axis placement of all the models on less objective attributes, like wordiness, willingness to give up, an ability to "think ahead" and pre-solve possible problems in code, for example, that I didn't think of or didn't think of talking about, etc etc etc.
For example I've been really enjoying Deepseek v4.1 Flash, it's very "straightforward" to the point of being almost dumb sometimes, but it's absolutely relentless and would solve almost any problem no matter how inefficient the solution is.
No idea how to measure all that, just average CoT length per task is probably a good approximation for some things, but not others.
1. In real life, most of us use token packages like OpenCode Go etc.
It would be handy to have a site like this one that takes into account the various deals and attempts to calculate the number of tokens per monthly fee for a chosen model. I realize this makes the task a lot more difficult.
2. It would be handy to have a chart like that for the AI hardware that people own. It helps you decide which model to run (resulting in different levels of intelligence and speed). Also difficult to please everyone (preprocessing vs token generation for example) and to keep updated!
I found https://llm-list.com/ yesterday and when I had a detailed look, I quickly found outdated entries, for example looking at GLM 5.3 flash it listed several providers as "free" that weren't free any longer.
Some of the charts are not uptodate.
For example "Raw capability: Math Index", shows GPT-5.2 as highest. It does not even mention astra, of fable.
More benchmark noise. They’re already problematic per se, and the error bars only get bigger by aggregating and adding costs on top of it.
Create your own, private evaluation system that reflects your use and your constraints. Focus on hard cases you encountered, find cases that break on one tier and not on another.
Cost per token is an extremely naive way to index cost, renders this chart essentially meaningless.
I wrote (i.e. let AI write) something like this for Opencode Go: https://6bj.de/aaogo/go_value_report.html
It compares Artificial Analysis scores against usage limits on Opencode Go, so I can see where to waste my quota most efficiently. Updated whenever I feel like it.
I guess it’s hard to actually calculate, but most people are probably on subscription plans with Anthropic or OpenAI and there’s not any good resources to estimate which models are the most “usage limit” efficient. I suppose you can sort of correlate raw API costs with how much usage a model would probably drain, but it would be nice if there was a reliable place to get that info.
This definition of cost is not particularly useful. You want cost per task, not cost per 1m tokens. Artificial Analysis does a good job of this.
> "$0.23 to under $0.24 - Qwen3.8-Flash-Next"
There are only 6 categories in the site, but a category entirely created out of thin air for Qwen3.8-Flash-Next?
Literally "$0.23 to under $0.24" means EXACTLY $0.23.
Is there a cheaper model than Gemini 3.8 Flash (High) that maybe/kind-of is on-par with it? For me it works really good but hit the limit in two hours tops... last week was the first time I hit the weekly limit and had to wait 4 days... Claude patches OK, but that is also getting drained really fast these days...
Coding and math graphs are very interesting. Extremely cheap models make it into the upper echelon, delivering 90% of the performance for 1% of the price compared to the #1.
What's impeding a lab from releasing its own new model, pricing it really low for the beautiful Pareto plot, accompanied by phrases VC love like "establishing a new frontier in cost", to just then raise prices back up?
The coding and math tabs seem to be missing the latest models...
Fun, but this seems to assume all LLM run on SAAS subscriptions. I would like to compare this to local LLM costs by converting my usage load x hardware costs into a token price. In addition if it games the LLM so often, these eventually are optimized and basically cheat on the score.
I won’t touch Meta models. I simply don’t trust or like Zuck and especially Wang
Updated daily and yet the frontier for math is listed as gpt-5.2
It's missing Opus 5.5 which was released over a day ago (and also is clearly on the pareto frontier).
I find it interesting that with the given metric comparison, for coding at min 50 strength, every frontier model brand is from a distinct vendor: Ling, Qwen, Gemini, Muse, Grok, GPT, and Claude in increasing value.
Interested to see the Coding and Math scores for Grok 4.7 when they’re available. Grok 4.6 is apparently already on the frontier for Coding.
If you maintain this over time, maybe include other sources than just AA and update the design to look less like zero-shot claude styling (I know that font! I know that color! Lol) it's genuinely useful :)
all I ever wanted is an updated website where I can see the best models I can run on my different devices locally, I don't get why people are throwing money at these companies
I’d love this but for every hardware
The page looks Claude generated and ranks Opus at the top.
Astra worse than 5.6 Sol is crazy work
I'm sure that local LLM will be far cheaper
I tried to build a very similar “frontier” [1] for myself and also include writing (which is highly debatable)
[1] https://philippdubach.com/posts/jev-model-router-for-pi/
When a clanker is slopping up a webpage, why does it have to be #f4f3ee? Curious.
Luna at the bottom of intelligence is laughable...
[dead]
[flagged]
You're better off going directly to artificial analysis, this is a feature-poor/misleading/outdated repackaging
Ex. this type of price estimation is quite naive - some models can require 2-3x the number of tokens to achieve the same level of intelligence. Artificial Analysis' own cost per task is a more fair estimation of cost.