75 comments

[ 1.8 ms ] story [ 64.9 ms ] thread
Their dinner plate chips are impressive.
GPT 5.6 Luna Ultrafast when?
The corresponding OpenAI post https://openai.com/index/previewing-ultrafast/

There is no pricing info, which could mean it's "if you have to ask..." territory or they are simply gauging interest before deciding

They're nearly certainly going to use it internally to speed up research that is serially bottlenecked. I would bet this is why they're interested in the Cerebras partnership more than everything else
i would assume internally they have even better tps
> Compared with output speeds reported by Artificial Analysis GPT-5.6 Sol on Ultrafast mode runs 11x faster than Fable 5, and 5x faster than Opus 4.8 on Fast mode.

Awesome work. I'm personally very excited for faster models/inference.

I think speed is underrated to some degree in the current conversation. For a while, I was using Cursor's Composer quite a lot, even over frontier models, just because of how darn fast it was.

I guess Gemini 3.7 Flash is no longer at the pareto frontier of speed to intelligence.
This is really cool. Someone here commented about similarity between this and hardware advancements for AV encode/decode.

I think it's only a matter of time before miniaturization can have a thumbnail sized user-replaceable accessory that contains the LLM built onto the hardware. I admit I don't know how any of that works, but would be amazing to experience. Fully local, fully offline, ultra fast local inference better than any personal computing product.

In 5-10 years nice smartphones will be able to run ChatGPT (~gpt3-4) class models. A memory rich laptop (highend mac/framework) can run GPT-OSS:120b or full Gemma4 at very interactive speeds.

High end phones can already run the smaller models at enough speed to be probably useful, especially for background/overnight photo tagging and curation and things like that.

I've been waiting so long for something amazing to come out of the OpenAI and Cerebras collaboration.

> In our evaluations, GPT-5.6 Sol on Ultrafast mode answered all 2,500 HLE questions in 11 hours and 11 minutes. Claude Fable 5 needed 78 hours and 27 minutes, more than three days of continuous compute, to arrive at the same conclusions. In other words, Ultrafast worked through the frontier of human knowledge in a single working day, achieving comparable accuracy nearly 7× faster.

This is actually insane.

Hopefully the release ultrafast of Terra and Luna too.

Seems like they will do Sol first while capacity constrained? I can't imagine the margins they'll be charging
Answering 2,500 independent questions is an embarrassingly parallel workload, all it needs is scale out. It would be more meaningful to know how much time was required for a single complete answer to a difficult HLE question.
Output from Cerebras with GPT model is 750 tokens per second.

Don’t blink.

(Chatjimmy has 14,200 TPS.)

An irrational gripe of mine is how GPT uses 7× instead of 7x.

I recognize that the former is the multiplication symbol, but I don't think it should be used that way.

I'm finding Luna suprisingly adequate for my work. I slept on it due to the benchmarks, but it's very fast and even on low reasoning I'm finding it more than adequate for "menial" work. (The speed is crucial for "interactive" work -- if a model is fast enough it goes from "async" to "real time", subjectively, which is a huge difference.)

In fact, I'd say it's overqualified for the kind of work I'm doing, because it spends >half the time verifying trivial changes (and the verification isn't as helpful as you'd expect, even with bigger models).

Maybe I can prompt it to be less aggressive about that (the new GPT models do it even without prompting).

Anyway, Ultrafast Luna would be amazing, though I strongly doubt they can offer Cerebras at anything approaching the current prices. Now we wait for Moore's Law? :)

I discovered yesterday that the “amazing thing that comes out of OpenAI” is Sol, due to its token efficiency.

Dollar for tokens, Sol and Fable are the same price.

However, Sol uses (literally: in testing) around 10-100x less output tokens compared to Fable for the same task.

We run our frontier models nearly 24/7, so switching to Sol will save us around $500 per day.

And, due to less guardrails, Sol also performed better, and we lost less tokens due to guardrails shutting down sessions (I feel like it’s illegal to take $50 of someone’s token money and then shut down a session with guardrails before they get an answer, and yet Anthropic do it to us constantly… either take our money and commit, or trigger the guardrails immediately)

Isn't ultrafast just making hundreds of subagents?
Wow, that's even faster than diffusion LLMs but with the Fable-level quality! Congrats!
I swear that now frontier AI stuff comes out few times a week.
This kills the crab.

Compilation time will be a genuine bottleneck for slop coding if this becomes the standard generation rate over the next few years. Go, Zig or even C99 with TCC for dev builds, any language that can get you systems-level performance (or close to it) in a dev environment where you can iterate in ms rather than minutes is going to be immensely more appealing than generating a potential prototype in 10 seconds and waiting 15 minutes for it to compile.

I think it's pretty obvious that, in that world, the AIs will simply be tasked with making the compilers faster. It's already happening with their own stack, after all.
Fast mode is already 1.5 times faster and 2x more expensive in the Codex subscription plan. If this thing is 14 times faster, then I can imagine running out of my quota in one session.
There is zero chance this will be offered to subscription users.
> allowing Sol Ultrafast to accelerate your most time-sensitive, mission-critical work

Curious, what are some of the use cases?

Agents would be excellent 'on call' team members. able to log in to production and diagnose issues, fix or raise issues, post outages etc.
I'd just like to point out that the largest model Cerebras has ever served is Kimi K2.6 which is 1T parameters, so that either means that theyve had a breakthrough on the hardware engineering side of things, or GPT-5.6 Sol is likely a lot smaller than people think.

If it truly is only ~1-2T parameters, then this kinda kills 2 narratives for me.

1. all the handwringing about open source catching up via Kimi K3 (3T params) is complete nonsense. All that matters imo for determining which labs are leading is intelligence per parameter. Anyone with a enough compute can train a giant model, but being able to squeeze capabilities into smaller models gives you a massive inference and training edge.

2. Inference margins are clearly insane, and this explains why OpenAI was able to lower the price of Luna by 80%. Id guess that thing is probably 120b params based on the TPS they are serving it at.

Would be extremely interesting if some of the closed models would be that small. Means maybe in future they could run locally.
Don't frontier labs distill their own bigger models into smaller ones? Opus 5 was probably distilled from Fable/Mythos. Since Chinese labs now have competitive models, they can distill those into their smaller version in Kimi K3.1 or something and achieve better intelligence per parameter results.
Whoa. This looks both powerful and expensive.

My prediction is that, this time next year, top developers outside ai labs will be spending 50k USD+ on inference.

Within labs, I've heard spend is already far beyond this per developer.

50k per month?!

If someone subsidize maybe, but if the companies need to pay no way, unless there is hard evidence of the return.

not sure how "top developers" are defined here, but there is huge diminishing return curve starts kicking in after $200/month price point for typical eng work.
> top developers outside ai labs will be spending 50k USD+ on inference

I think it is more like top companies, not top developers, and the problem with developers in top companies was and is - absolute majority of them are not actually directly working on things that increase revenue, so companies can spend a ton of money and see barely if any changes in the product and the bottom line, so companies, at least legacy ones will be reluctant to sponsor that long term.

Isn't TCO lower with Cerebras chips compared to Nvidia? Theoretically, most developers should eventually be running on Ultrafast.
Unless I have read over it, besides the animation in the intelligence vs speed graph which only mentions internal data and not whether they truly reran the AA suite, there is no actually solid statement on the important aspect of performance.

Neither the Cerebras or OpenAI post [0] outright state that this performs exactly the same as regular 5.6 Sol. I feel if this was 1:1 just Sol but much faster, they'd (rightfully) scream that off the rooftops. A line such as "this is the same performance, just faster, with no downsides" would go a long way in clarity and communication. Along with no pricing information, I'll hold out on further information.

[0] https://openai.com/index/previewing-ultrafast/

"delivering up to 750 output tokens per second and without any quality compromise" seems pretty definitive.
This is what Cerebras does- take other people's models and run them very very fast.
You (or anyone else) can just benchmark and compare. If they were serving a dumber model it would be trivially detectable.
I mean I still regularly use 5.3 spark (the cerebrus model) that comes with my sub to do rapid reviews of 5.6's work and it finds oodles of problems in about a minute.
Meanwhile they are down 12,68% today because of disappointing earnings.
This is something I'm ready to pay for. Not more per token, but I will be happy to burn through 20x Pro subscription as fast as I consume my Plus weekly limit now, with 10x more tokens per unit of time. I've learned how to deal with and steer Sol medium quite efficiently, but at the same time I realize it's so slow for the small tasks it can do well, and still so unreliable for open-ended tasks.
I haven't wrapped my head around what level of reasoning this involves. Is it equivalent to max?

I didn't like Sol initially but it is growing on me the more I use it. Its personality is a bit flat and I caught it taking shortcuts a few times. But once I learned how to interact with it, I'm genuinely warming up to it. I find that it writes code that has fewer bugs even than Fable (although, to be fair I reach for Fable when the task is less well defined).

If this has similar performance to Sol at max reasoning level, this would be a compelling reason to shift even more of my work (maybe the majority) to this model.

Did Cerebras get rid of their like $1500 per month plans for open models?
“our stock price went down today, here’s something to feed it”
> GPT-5.6 Sol on Ultrafast mode, delivering up to 750 output tokens per second

https://taalas.com/products/

> delivering 17k tokens per second per user on Llama 3.1 8B model.

Obviously this is a much smaller model, but I really can't wait for ASICs to take over the LLM space.

Imagine running a model like Sol/Fable (even half the size with 60-70% of it's intelligence) on your own ASIC hardware.

Good news for Intel and AMD.

Rught now on large scale codebases the bottleneck is both claude/codex inference, as well as time it takes to run tens of thousands of tests. We put those workloads on dedicated epyc 9005 build machines - but it still takes minutes per run. Those who can afford the fast tokens will be in the market for faster CPU that money can buy today.