LFM2.5 24B-A2B on NVIDIA RTX PRO 6000 Blackwell

Can the NVIDIA RTX PRO 6000 Blackwell run LFM2.5 24B-A2B?

Yes, it runs great

Runs great. At F16 it needs about 53.5 GB of 96.0 GB usable, with an estimated speed of 110 tok/s.

Best quant: F16
Generation speed: 110 tok/s
Prompt processing: 36,735 tok/s
First token (1K prompt): 0.3 s
Largest safe context: 128K

Best app to run it on the NVIDIA RTX PRO 6000 Blackwell

On Windows

Ollama

Command line + local API · runs on llama.cpp (ggml), MLX (Apple silicon, preview)

One command with Ollama’s own build of this model.
Serves an OpenAI-compatible API that other apps and coding tools can use.

ollama run lfm2.5:24b-a2b-bf16

LM Studio Point-and-click: search, download and chat in one window.
llama.cpp Usually the first to support new model architectures.

On Linux

Ollama

Command line + local API · runs on llama.cpp (ggml), MLX (Apple silicon, preview)

One command with Ollama’s own build of this model.
Serves an OpenAI-compatible API that other apps and coding tools can use.

ollama run lfm2.5:24b-a2b-bf16

vLLM The full model fits in GPU memory at 8-bit, so it can batch many requests: best for serving a team or an app.
LM Studio Point-and-click: search, download and chat in one window.

Want an agent? LFM2.5 24B-A2B supports tool use, so you can pair a llama.cpp or MLX LM server with one of these. Apps that run on this computer come first.

Coding agents

Personal assistants

How we choose: the files published for this model, your operating system and GPU, whether it fits fully in memory, and published speed tests (MLX vs llama.cpp on Apple silicon, vLLM vs llama.cpp vs Ollama on NVIDIA). Details.

Download LFM2.5 24B-A2B

Recommended for the NVIDIA RTX PRO 6000 Blackwell: F16. Every other quantization has its own download link in the table below.

Quickest · Ollama

One command downloads a ready-to-run version and starts a chat.

Get Ollama at ollama.com/download (opens in a new tab) · model page (opens in a new tab)

ollama run lfm2.5:24b-a2b-bf16

Quantized files · GGUF

Works in LM Studio, llama.cpp, Jan and Ollama. No sign-in needed.

Download the F16 file (opens in a new tab)

huggingface.co/LiquidAI/LFM2.5-24B-A2B-GGUF (opens in a new tab)

ollama run hf.co/LiquidAI/LFM2.5-24B-A2B-GGUF:BF16

Original weights · Hugging Face

Full-precision files from Liquid AI, for fine-tuning, vLLM, Transformers.

huggingface.co/LiquidAI/LFM2.5-24B-A2B (opens in a new tab)

No sign-in needed.

Official repo · GitHub

Reference code, inference scripts and docs from Liquid AI. The weights themselves download from Hugging Face.

github.com/Liquid4All/cookbook (opens in a new tab)

Liquid AI’s official cookbook for running LFM models.

git clone https://github.com/Liquid4All/cookbook

Links checked 1 day ago against Hugging Face, Ollama and GitHub.

How it’s rated

Quality 74Great quality (A) · 74 of 100: how capable the model is at F16, whatever it runs on. Match score 86/100 (quality, fit, speed and memory use) sets the ranking.: How capable the model itself is at F16, from its editorial quality tier scaled by the quantization or precision. It’s the same on any hardware that runs this variant. Scale: A+ 85+ · A 70+ · B 55+ · C 40+ · D under 40.
Runs Runs great: How well it runs on the NVIDIA RTX PRO 6000 Blackwell: speed, capped by how well it fits in memory. Runs great: 30+ tokens/s (or near the image, video or decision speed target) · Runs well: 10+ tokens/s · Slow: 3+ tokens/s · Very slow: under 3 · Won’t run: doesn’t fit even with offload. A tight fit or RAM offload caps it at Runs well; heavy offload caps it at Slow.

Match score

To order lists, each result also gets a match score out of 100 from four parts. It favors the most capable model that still runs well.

Model qualityEditorial quality tier of the model, scaled by the quantization or precision used.	33.3 / 45
Memory fitFull marks when everything sits in GPU memory; fewer when part spills to system RAM.	25.0 / 25
SpeedGeneration speed against a 40 tokens/s target for LLMs, or render time against a target for image and video.	15.0 / 15
Memory useRewards using the memory you have, so a bigger, better variant beats a tiny one that leaves most memory idle.	12.9 / 15
Match score	86.2 / 100

The match score orders results; it is not shown as a grade. Quality and Runs are the two ratings in the tables.

Every quantization on the NVIDIA RTX PRO 6000 Blackwell

Quant	Memory needed	Available	Speed	Runs	Quality	Download
`Q2_K`	15.2 GB	96.0 GB	263 tok/s	Runs great	51Fair quality (C) · 51 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking.	GGUF files (opens in a new tab)
`Q3_K_M`	17.3 GB	96.0 GB	244 tok/s	Runs great	62Good quality (B) · 62 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking.	GGUF files (opens in a new tab)
`Q4_K_M`	20.2 GB	96.0 GB	223 tok/s	Runs great	70Great quality (A) · 70 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking.	GGUF (opens in a new tab)
`Q5_K_M`	22.6 GB	96.0 GB	206 tok/s	Runs great	71Great quality (A) · 71 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 77/100 (quality, fit, speed and memory use) sets the ranking.	GGUF (opens in a new tab)
`Q6_K`	25.2 GB	96.0 GB	192 tok/s	Runs great	73Great quality (A) · 73 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 79/100 (quality, fit, speed and memory use) sets the ranking.	GGUF (opens in a new tab)
`Q8_0`	31.0 GB	96.0 GB	167 tok/s	Runs great	74Great quality (A) · 74 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking.	GGUF (opens in a new tab)
`F16`	53.5 GB	96.0 GB	110 tok/s	Runs great	74Great quality (A) · 74 of 100: how capable the model is at F16, whatever it runs on. Match score 86/100 (quality, fit, speed and memory use) sets the ranking.	GGUF (opens in a new tab) Original weights (opens in a new tab)

Memory figures assume 8K of context. Longer conversations or documents grow the KV cache.

Other hardware that runs it

74Great quality (A) · 74 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 88/100 (quality, fit, speed and memory use) sets the ranking. Apple M4 Pro (48 GB) Q8_0 · 48.8 tok/s · Runs great
73Great quality (A) · 73 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA GeForce RTX 5090 Q6_K · 192 tok/s · Runs great
74Great quality (A) · 74 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. Apple M1 Max (64 GB) Q8_0 · 65.8 tok/s · Runs great
74Great quality (A) · 74 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. Apple M4 Max (64 GB) Q8_0 · 82.4 tok/s · Runs great
71Great quality (A) · 71 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. AMD Radeon RX 7900 XTX Q5_K_M · 156 tok/s · Runs great
71Great quality (A) · 71 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA GeForce RTX 3090 Q5_K_M · 164 tok/s · Runs great

Estimates come from published hardware specs and our open formulas, not from running the model on your machine. Real speed varies with runtime, drivers and settings. Numbers marked “Measured” come from a linked benchmark.