LFM2.5-24B-A2B: Built for fast on-device agents

> it runs up to 1.7× faster than comparable MoEs on-device, with 184 tokens/s on an Apple M5 Max and up to 21K output tokens/s on a single NVIDIA B300 GPU

Wow, those are seriously impressive numbers for how intelligent this model seems to be!

One thing that this article doesn't clarify, and that I guess I really should look into, is what exactly AutomationBench tests for as compared to other benchmarks. The standout performance on AutomationBench is interesting to see, but I don't know what it really means because the model seems to score relatively close to other local models on every other benchmark.

It is very fast on a 4090, but also very stupid.

Which makes for a dangerous combination, because it will delete prod, and it will do so at 250t/s.

Very cool. I’ve been on Gemma 4 12B because of my limited RAM. There’s already an Unsloth quantization of this model so I’m downloading that and will check its performance in the morning. Someone else in this thread recommended Ternary-Bonsai-27B, which I’ll also try.

Just also wanted to note that this model has a questionably-enforceable non-commercial use clause in its license. I really wish they had consulted with an IP lawyer, it’s very confusing.

In a private benchmark I have, it performs much, much worse than Qwen 35B A3B for general knowledge. Still, probably beats Qwen 9B at much higher speed.

I asked it to implement the simplest todo in ghostty codebase and it failed badly - kept going in circles fixing one compile error and introducing another. Qwen 3.8 was slower (40 tok/s compared to 75) but did get it done.

This seems like a big deal, no?

If Qwen had released a Qwen3.8 MoE of similar size, with a naive extrapolation its capabilities would roughly be similar to this model. So basically someone has caught up to Qwen (albeit Qwen from 1-2 months ago) in parameter efficiency.

Maybe someone knows what order of cost a release like this requires in real terms of hardware and energy spend?

The cost over the diff from the moment of beginning training to the end. It's iterative from a prior model so it starts there and becomes this release after spending occurs I presume.

For example, is liquid spending millions in energy in this case and the cost is amortized over successive training runs over the initial datacenter spend? What if it's just eating cloud costs and has no dedicated hardware, that sort of thing. I don't particularly mean this model and liquid specifically. Just for this class of smaller fast on-device range of model what is the order of the training cost in real terms?

Maybe its short sighted to ask it as if it's a singular training run and is perhaps the end state of many variations of testing thus all of the R&D training runs would have to be included to to inform what became the final release. Seems like this should be taken into consideration. If anyone happens to know it'd be appreciated!

IANAL ... but how does Commercial Use Threshold applies if I decide to use this at work to make my life easier as a local mlx runtime and not to deploy it and make it available to end users, customers, and in products built by the company?

Is that a trigger or a grey area? The company will profit from my work, and my work would be empowered by this model, is that a "use for commercial purposes" or not?

i'm gettin between 200-320 tok/s on a 5070ti with 131k context on llamacpp and no tuning.

very nice.

I can't speak to the intelligence of it yet.

these small agentic-focused models are a lot of fun to play with if you haven't yet.

Love seeing the effort Liquid is putting into local models. I’m looking forward to a lot of progress here in the next 5 years

I'm running a Ryzen 7 8745HS with Radeon 780M (27GB RAM), using Ollama's Vulkan backend.

I get ~45 tokens/s generation and ~680 tokens/s prompt processing.

The problem is ollama is not caching the prompt, I get -> forcing full prompt re-processing due to lack of cache data

So this wouldn't work in its current state in an agentic loop.

Tried using it yesterday via OpenRouter in OpenCode to follow a plan written by a larger model. I didnt find it to be particularly great. It got stuck on a lot of tool calls. Given the price and speed, it would be nice if it could do what it advertises.

Perhaps there are some system prompt tweaks that could get it in line

I don't quite understand how people use these kind of local models that are specifically trained for agentic use cases. Like, what do they actually do with them? Do you use them as an assistant like with OpenClaw?

When used for coding, how do you divide the work between agents? Use a relatively fast model, like this one, as the main agent that calls out to other subagents that write the actual code? But then you'd need multiple models loaded, and why not just use the coding model to do everything? Do you use hosted cloud models for coding and code review, and this local one for orchestration? I don't see the benefit of that workflow either.

I've used local models to write some scripts etc, I like the idea of local computation instead of buying tokens from the cloud, but I don't know what the various agentic models are for, if not code generation.

> with 184 tokens/s on an Apple M5 Max

That seems pretty decent. How many tokens/s do you need for a coding agent to feel like Claude or codex?

I could see it being useful for say summarising content and similar, but speed increases like this are rarely free lunches.

not sure why this is being compared to 2-generations old versions of qwen

Does anyone know the order of magnitude licensing cost for the model for businesses over 10m in revenue?

The license is generrous (free under <$10m) Pricing confusing when it shows API pricing 1st

Always good to hear for on-device suited model, but always cannot sure how it real-performance in real task..

It is a good model for on-device work.

[flagged]

[flagged]

[dead]

[flagged]

[dead]