RunsOnMy ModelsGPUs & MacsBy VRAMAPI pricesVRAM calculatorNew

Can LFM2.5 24B A2B run on the RTX 5090?

Yes. At 8K context the best fit is Q8_0, using about 24.5 GB of the 30.4 GB usable on the RTX 5090. Estimated speed ceiling: ~843 tokens/sec.

Memory by quant and context

Quant4K ctx8K ctx32K ctx128K ctxMax tok/s (est., 8K)
2-bit (Q2_K)9.1 GB Fits9.2 GB Fits9.6 GB Fits11.5 GB Fits2389
3-bit (Q3_K_M)11.6 GB Fits11.7 GB Fits12.2 GB Fits14.1 GB Fits1833
4-bit (Q4_K_M) •14.3 GB Fits14.3 GB Fits14.8 GB Fits16.7 GB Fits1479
5-bit (Q5_K_M) •16.6 GB Fits16.7 GB Fits17.1 GB Fits19.0 GB Fits1261
6-bit (Q6_K) •19.1 GB Fits19.1 GB Fits19.6 GB Fits21.5 GB Fits1091
8-bit (Q8_0) •24.5 GB Fits24.5 GB Fits25.0 GB Fits26.9 GB Fits843
16-bit (F16/BF16) •90.1 GB No90.1 GB No90.6 GB No92.5 GB No—

• real GGUF file size from LiquidAI/LFM2.5-24B-A2B-GGUF. Speed is a bandwidth ceiling (1792 GB/s), not a benchmark. Methodology.

More

LFM2.5 24B A2B on every GPU and Mac · Every model that runs on the RTX 5090 · Other models that fit 32 GB VRAM · Try other settings in the calculator

Data: Hugging Face Hub (LiquidAI/LFM2.5-24B-A2B), NVIDIA specs. Data as of 2026-11-17.