Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Running Gemma 4 26B at 5 tokens/sec on a 13-year-old Xeon with no GPU

neomindlabs.com

Post reply on HN
Print view View original

51–60 of 227 posts

Re: Running Gemma 4 26B at 5 tokens/sec on a 13-year-old Xeon with no GPU

#51

Post by smeeth »

dwa3592 wrote: post #21

I have a prediction. By the mid of 2027, we will have >200B MoE models running on basic consumer hardware. I am running Qwen3.6-35B-A3B locally on my 16GB mac with 7-9 tokens/second. Link - https://github.com/deepanwadhwa/samosa-chat This is a GPT4 level model running locally with a decent speed on a 16gb ram macbook air.

Probably won't have to wait that long. Prism released Bonsai 27B (https://huggingface.co/prism-ml/Ternary-Bonsai-27B-mlx-2bit) as a ternary model a few days ago, its just ~7GB and runs at 44+ t/sec on an m4 max laptop. That's already in the ballpark of active parameter count of most 200B+ models, so we will get a model like this whenever Prism feels like releasing one.

It is debatable if we will actually need that many parameters though, since recursive nets like HRM (https://huggingface.co/sapientinc/HRM-Text-1B) don't need to parametrize as heavily.

Re: Running Gemma 4 26B at 5 tokens/sec on a 13-year-old Xeon with no GPU

#52

Post by dwa3592 »

rvz wrote: post #43
dwa3592 wrote: post #21

I have a prediction. By the mid of 2027, we will have >200B MoE models running on basic consumer hardware. I am running Qwen3.6-35B-A3B locally on my 16GB mac with 7-9 tokens/second. Link - https://github.com/deepanwadhwa/samosa-chat This is a GPT4 level model running locally with a decent speed on a 16gb ram macbook air.

> I am running Qwen3.6-35B-A3B locally on my 16GB mac with 7-9 tokens/second. That is no where near decent at all.

it's a 16GB machine. i am proud of this machine so far.

Re: Running Gemma 4 26B at 5 tokens/sec on a 13-year-old Xeon with no GPU

#54

Post by dwa3592 »

smeeth wrote: post #51
dwa3592 wrote: post #21

I have a prediction. By the mid of 2027, we will have >200B MoE models running on basic consumer hardware. I am running Qwen3.6-35B-A3B locally on my 16GB mac with 7-9 tokens/second. Link - https://github.com/deepanwadhwa/samosa-chat This is a GPT4 level model running locally with a decent speed on a 16gb ram macbook air.

Probably won't have to wait that long. Prism released Bonsai 27B ( https://huggingface.co/prism-ml/Ternary-Bonsai-27B-mlx-2bit ) as a ternary model a few days ago, its just ~7GB and runs at 44+ t/sec on an m4 max laptop. That's already in the ballpark of active parameter count of most 200B+ models, so we will get a model like this whenever Prism feels like releasing one. It is debatable if we will actually need that…

agreed!! in my heart i really wanted to say by the end of 2026 but wanted to add some wiggle room in case they start to ban open source AI development.

Re: Running Gemma 4 26B at 5 tokens/sec on a 13-year-old Xeon with no GPU

#55

Post by notnullorvoid »

dwa3592 wrote: post #21

I have a prediction. By the mid of 2027, we will have >200B MoE models running on basic consumer hardware. I am running Qwen3.6-35B-A3B locally on my 16GB mac with 7-9 tokens/second. Link - https://github.com/deepanwadhwa/samosa-chat This is a GPT4 level model running locally with a decent speed on a 16gb ram macbook air.

I mostly agree with the prediction though maybe a bit more pessimistic about the timeline. Also I'm not sure our current usage of parameter count would make sense in this scenario, such a feat would require compressing current parameters in a manner much different then something producing a bit count per parameter. A hypothetical example would be using a single seed parameter per layer which then passed into a noise function produces the functional weights for that layer, able to reduce per weight size to sub bit levels (256 bit seed, producing 16K weights).

Re: Running Gemma 4 26B at 5 tokens/sec on a 13-year-old Xeon with no GPU

#56

Post by ctoth »

LeifCarrotson wrote: post #38
reidrac wrote: post #27

Earlier quoted context omitted.

> on a decent speed But you said 7-9 tokens/second , that's not a decent speed. I'm not an expert by all means but in my local experiments, less than 12 to 16 tps is too slow.

I think that any workflow that requires the user to stare at the tokens being generated live is using it wrong. Delegate, don't stare! https://mikeveerman.github.io/tokenspeed/?rate=10&mode=text You think of an idea that you want to have the LLM process, queue it up, and go back to what you were doing. Once you've finished reading the next article on HN about a 5 tps Xeon, your task will be complete. It's kind of lik…

[dead]

Re: Running Gemma 4 26B at 5 tokens/sec on a 13-year-old Xeon with no GPU

#59

Post by Azantys »

Capricorn2481 wrote: post #46
dwa3592 wrote: post #21

I have a prediction. By the mid of 2027, we will have >200B MoE models running on basic consumer hardware. I am running Qwen3.6-35B-A3B locally on my 16GB mac with 7-9 tokens/second. Link - https://github.com/deepanwadhwa/samosa-chat This is a GPT4 level model running locally with a decent speed on a 16gb ram macbook air.

I tried Qwen3.6-35B-A3B, but it couldn't generate a 50-100 line Clojure file without having broken parens mismatches. I know Clojure isn't super popular, but the syntax is pretty simple and the frontier models do fine with it.

You are comparing a 35B models to a 635B+ frontier model, of course thats not even close

Re: Running Gemma 4 26B at 5 tokens/sec on a 13-year-old Xeon with no GPU

#60

Post by allknowingfrog »

LeifCarrotson wrote: post #38
reidrac wrote: post #27

Earlier quoted context omitted.

> on a decent speed But you said 7-9 tokens/second , that's not a decent speed. I'm not an expert by all means but in my local experiments, less than 12 to 16 tps is too slow.

I think that any workflow that requires the user to stare at the tokens being generated live is using it wrong. Delegate, don't stare! https://mikeveerman.github.io/tokenspeed/?rate=10&mode=text You think of an idea that you want to have the LLM process, queue it up, and go back to what you were doing. Once you've finished reading the next article on HN about a 5 tps Xeon, your task will be complete. It's kind of lik…

We clearly have different goals. I want an LLM to review my code, not the other way around.
Post reply on HN