LFM2 · 24B · MoE
Can you run LFM2 24B-A2B Instruct locally?
LFM2 24B-A2B Instruct is a mixture-of-experts model with 24B parameters, 2B active per token. At 4-bit or better it runs on 85 of the 120 hardware configurations listed here; the smallest is the RX 7900 XT (20 GB). On a 128 GB DGX Spark it runs at Q4_K_M, at roughly 129 tokens per second (our est.).
Data as of 2 November 2026
Fastest machines for LFM2 24B-A2B Instruct
Every machine that runs it, fastest first. Filter by type; your machine is marked. Compare machines running it side by side
- 1RTX 5090Graphics card · 32 GB~366 tok/s ModelFit est.Q4_K_M · 15.3 of 31 GB
- 2RTX PRO 6000 BlackwellGraphics card · 96 GB~366 tok/s ModelFit est.Q4_K_M · 15.3 of 95 GB
- 32x RTX 5090Graphics card · 64 GB~330 tok/s ModelFit est.Q4_K_M · 15.3 of 62 GB
- 42x RTX PRO 6000 BlackwellGraphics card · 192 GB~330 tok/s ModelFit est.Q4_K_M · 15.3 of 190 GB
- 5RTX 4090Graphics card · 24 GB~257 tok/s ModelFit est.Q4_K_M · 15.3 of 23 GB
- 6M5 UltraMac · 96–512 GB · 5 variants~239 tok/s ModelFit est.Q4_K_M · 15.3 GB
- 72x RTX 4090Graphics card · 48 GB~239 tok/s ModelFit est.Q4_K_M · 15.3 of 46 GB
- 8RTX 3090 TiGraphics card · 24 GB~236 tok/s ModelFit est.Q4_K_M · 15.3 of 23 GB
- 9RTX 3090Graphics card · 24 GB~225 tok/s ModelFit est.Q4_K_M · 15.3 of 23 GB
- 102x RTX 3090 TiGraphics card · 48 GB~220 tok/s ModelFit est.Q4_K_M · 15.3 of 46 GB
- 112x RTX 3090Graphics card · 48 GB~211 tok/s ModelFit est.Q4_K_M · 15.3 of 46 GB
- 12M5 MaxMac · 48–128 GB · 3 variants~178 tok/s ModelFit est.Q4_K_M · 15.3 GB
- 13RX 7900 XTXGraphics card · 24 GB~176 tok/s ModelFit est.Q4_K_M · 15.3 of 23 GB
- 14M3 UltraMac · 96–512 GB · 5 variants~172 tok/s ModelFit est.Q4_K_M · 15.3 GB
- 152x RX 7900 XTXGraphics card · 48 GB~167 tok/s ModelFit est.Q4_K_M · 15.3 of 46 GB
- 16RX 7900 XTGraphics card · 20 GB~160 tok/s ModelFit est.Q4_K_M · 15.3 of 19.0 GB
- 17M4 MaxMac · 48–128 GB · 3 variants~149 tok/s ModelFit est.Q4_K_M · 15.3 GB
- 18M2 UltraMac · 64–192 GB · 7 variants~148 tok/s ModelFit est.Q4_K_M · 15.3 GB
- 19M5 Max · 36 GB · 32cMac · 36 GB~141 tok/s ModelFit est.Q4_K_M · 15.3 of 24 GB
- 20M1 UltraMac · 64–128 GB · 4 variants~131 tok/s ModelFit est.Q4_K_M · 15.3 GB
- 21DGX Spark (GB10)AI mini PC · 128 GB~129 tok/s our est.Q4_K_M · 15.3 of 122 GB
- 22Ryzen AI Max+ 395 · 64 GBAI mini PC · 64 GB~121 tok/s our est.Q4_K_M · 15.3 of 48 GB
- 23M3 MaxMac · 48–128 GB · 3 variants~119 tok/s ModelFit est.Q4_K_M · 15.3 GB
- 24M4 Max · 36 GB · 32cMac · 36 GB~118 tok/s ModelFit est.Q4_K_M · 15.3 of 24 GB
- 25M2 MaxMac · 32–96 GB · 5 variants~110 tok/s ModelFit est.Q4_K_M · 15.3 GB
- 26M5 ProMac · 24–64 GB · 6 variants~109 tok/s ModelFit est.Q4_K_M · 15.3 GB
- 27M1 MaxMac · 32–64 GB · 4 variants~98 tok/s ModelFit est.Q4_K_M · 15.3 GB
- 28M3 MaxMac · 36–96 GB · 2 variants~94 tok/s ModelFit est.Q4_K_M · 15.3 GB
- 29M4 ProMac · 24–64 GB · 6 variants~92 tok/s ModelFit est.Q4_K_M · 15.3 GB
- 30Ryzen AI Max+ 395 · 128 GBAI mini PC · 128 GB~69 tok/s ModelFit est.Q4_K_M · 15.3 of 96 GB
- 31M2 Pro · 32 GBMac · 32 GB~67 tok/s ModelFit est.Q4_K_M · 15.3 of 21 GB
- 32M6Mac · 24–32 GB · 3 variants~66 tok/s ModelFit est.Q4_K_M · 15.3 GB
- 33M1 Pro · 32 GBMac · 32 GB~60 tok/s ModelFit est.Q4_K_M · 15.3 of 21 GB
- 34M5Mac · 24–32 GB · 4 variants~57 tok/s ModelFit est.Q4_K_M · 15.3 GB
- 35M3 Pro · 36 GBMac · 36 GB~55 tok/s ModelFit est.Q4_K_M · 15.3 of 24 GB
- 36M4Mac · 24–32 GB · 2 variants~42 tok/s ModelFit est.Q4_K_M · 15.3 GB
- 37M3 · 24 GBMac · 24 GB~37 tok/s ModelFit est.Q4_K_M · 15.3 of 16.0 GB
- 38M2 · 24 GBMac · 24 GB~35 tok/s ModelFit est.Q4_K_M · 15.3 of 16.0 GB
Too big for 35 machines
RTX 5080, RTX 5070 Ti, RTX 5060 Ti, RTX 4080 SUPER, RTX 4080, RTX 4070 Ti SUPER, RTX 4060 Ti, Tesla T4, Tesla P100, RX 9070 XT, RX 9060 XT, M3 Pro · 18 GB, RTX 5070, RTX 4070 Ti, RTX 4070 SUPER, RTX 4070, RTX 3060, M2 Pro · 16 GB, M1 Pro · 16 GB, M6 · 16 GB · 12c, M6 · 16 GB · 10c, M5 · 16 GB · 10c, M5 · 16 GB · 8c, M4 · 16 GB, M3 · 16 GB, M2 · 16 GB, M1 · 16 GB, RTX 5060, RTX 4060, RTX 3070, RTX 3060 Ti, RTX 2070, M3 · 8 GB, M2 · 8 GB, M1 · 8 GB. LFM2 24B-A2B Instruct needs about 15.3 GB at its smallest build.
Speeds are single-user decode estimates, not measurements. "ModelFit est." comes from ModelFit's model fitted on published runs (CC BY 4.0); "our est." is memory bandwidth divided by bytes read per token, times 0.55. Memory is what the best build needs, against what the machine can give a model. Method
About the model
- Parameters
- 24B
- Active per token
- 2B
- Type
- Mixture of experts
- Good for
- Local AI agents, privacy-first tool calling, MCP workflows
Builds and how to run them
| Quantization | Load | Min. RAM | Command |
|---|---|---|---|
| Q4_K_M | 14.0 GB | 24 GB | ollama run lfm2:24b-a2b |
Load, minimum RAM and Ollama commands: ModelFit, CC BY 4.0. Load is memory with the model loaded, before long contexts.
Questions
Can I run LFM2 24B-A2B Instruct on a Mac?
Yes on a Mac with enough memory: the smallest Apple configuration listed here that runs it at 4-bit or better is the M5 Pro · 24 GB · 20c.
How much memory does LFM2 24B-A2B Instruct need?
About 15.3 GB at the smallest build in our data, including a margin for the runtime and an 8K-token context. Longer contexts need more.
Does LFM2 24B-A2B Instruct run on a DGX Spark?
Yes, at Q4_K_M. The Spark has 128 GB of unified memory, about 121.6 GB usable.