Hardware requirements
What hardware do I need to run Fara 7B?
8.29B
125K context
MIT
At Q4_K_M the weights are 4.7 GB; with an 8K KV cache and runtime overhead that is 5.7 GB, so it wants a 8 GB card to run with headroom — in practice 8 GB VRAM: GeForce RTX 3070 Ti and 7 others. Every figure below is computed from the model’s published shape against each device’s usable memory; tokens per second are estimates. The method.
Minimum practical
8 GB VRAM
GeForce RTX 3070 Ti — Q5_K_M, 6.5 GB of 7.0 GB, ~67 tok/s, up to 16K context.
the smallest card that holds it at all, at Q3_K_M — the lowest quantisation worth running
Recommended
8 GB VRAM
GeForce RTX 3070 Ti — Q4_K_M, 5.7 GB of 7.0 GB, ~79 tok/s, up to 31K context.
runs the recommended quantisation with headroom at 8K context
High quality
16 GB VRAM
GeForce RTX 5080 — Q8_0, 9.2 GB of 14.4 GB, ~71 tok/s, up to 102K context.
near-lossless Q8_0 with headroom
Long context
16 GB VRAM
GeForce RTX 5080 — Q4_K_M, 12.1 GB of 14.4 GB, ~125 tok/s at 125K.
128K tokens (or the model’s maximum) at the recommended quantisation
Apple Silicon
16 GB unified
M1 · 16 GB — Q4_K_M, 5.7 GB of 10.7 GB, ~8.2 tok/s, up to 99K context.
smallest Mac that runs the recommended quantisation with headroom
CPU only
32 GB RAM
CPU only · DDR4 dual-channel — Q8_0, 9.2 GB of 25.6 GB, ~2.6 tok/s, up to 125K context.
no GPU — weights stream from system RAM
How much VRAM at each quantisation
| Quant | Weights | +KV 8K | +KV 32K | +KV 128K | Total @ 8K | Needs | Quality |
|---|---|---|---|---|---|---|---|
| F16 | 15.4 GB | 0.44 GB | 1.75 GB | 6.8 GB | 16.5 GB | 24 GB card | Reference |
| Q8_0 | 8.2 GB | 0.44 GB | 1.75 GB | 6.8 GB | 9.2 GB | 16 GB card | −0.1% ppl |
| Q6_K | 6.3 GB | 0.44 GB | 1.75 GB | 6.8 GB | 7.4 GB | 10 GB card | −0.4% ppl |
| Q5_K_M | 5.5 GB | 0.44 GB | 1.75 GB | 6.8 GB | 6.5 GB | 10 GB card | −0.8% ppl |
| Q4_K_M | 4.7 GB | 0.44 GB | 1.75 GB | 6.8 GB | 5.7 GB | 8 GB card | −1.9% ppl |
| Q3_K_M | 3.8 GB | 0.44 GB | 1.75 GB | 6.8 GB | 4.8 GB | 8 GB card | −5.4% ppl |
| Q2_K | 3.2 GB | 0.44 GB | 1.75 GB | 6.8 GB | 4.3 GB | 8 GB card | −15% ppl |
Totals add 0.6 GB runtime overhead and an f16 KV cache; "needs" is the smallest discrete card class whose usable memory (after display and driver reserve) leaves 15% headroom. A q8_0 cache roughly halves the KV columns.
NVIDIA · GeForce
| Device | Memory | Best quant with headroom | Fits at all | Max context @ Q4_K_M | Tok/s est. | 128K | |
|---|---|---|---|---|---|---|---|
| GeForce RTX 5090 | 32 GB | Q8_0 | Q8_0 | 125K | 233 | yes | Check |
| GeForce RTX 5080 | 16 GB | Q8_0 | Q8_0 | 125K | 125 | yes | Check |
| GeForce RTX 5070 Ti | 16 GB | Q8_0 | Q8_0 | 125K | 116 | yes | Check |
| GeForce RTX 5070 | 12 GB | Q6_K | Q8_0 | 97K | 87 | no | Check |
| GeForce RTX 5060 Ti 16 GB | 16 GB | Q8_0 | Q8_0 | 125K | 58 | yes | Check |
| GeForce RTX 5060 | 8 GB | Q4_K_M | Q5_K_M | 31K | 58 | no | Check |
| GeForce RTX 4090 | 24 GB | Q8_0 | Q8_0 | 125K | 131 | yes | Check |
| GeForce RTX 4080 Super | 16 GB | Q8_0 | Q8_0 | 125K | 96 | yes | Check |
| GeForce RTX 4080 | 16 GB | Q8_0 | Q8_0 | 125K | 93 | yes | Check |
| GeForce RTX 4070 Ti Super | 16 GB | Q8_0 | Q8_0 | 125K | 87 | yes | Check |
| GeForce RTX 4070 Ti | 12 GB | Q6_K | Q8_0 | 97K | 65 | no | Check |
| GeForce RTX 4070 Super | 12 GB | Q6_K | Q8_0 | 97K | 65 | no | Check |
| GeForce RTX 4070 | 12 GB | Q6_K | Q8_0 | 97K | 65 | no | Check |
| GeForce RTX 4060 Ti 16 GB | 16 GB | Q8_0 | Q8_0 | 125K | 37 | yes | Check |
| GeForce RTX 4060 Ti | 8 GB | Q4_K_M | Q5_K_M | 31K | 37 | no | Check |
| GeForce RTX 4060 | 8 GB | Q4_K_M | Q5_K_M | 31K | 35 | no | Check |
| GeForce RTX 3090 Ti | 24 GB | Q8_0 | Q8_0 | 125K | 131 | yes | Check |
| GeForce RTX 3090 | 24 GB | Q8_0 | Q8_0 | 125K | 122 | yes | Check |
| GeForce RTX 3080 Ti | 12 GB | Q6_K | Q8_0 | 97K | 118 | no | Check |
| GeForce RTX 3080 12 GB | 12 GB | Q6_K | Q8_0 | 97K | 118 | no | Check |
| GeForce RTX 3080 | 10 GB | Q6_K | Q6_K | 64K | 99 | no | Check |
| GeForce RTX 3070 Ti | 8 GB | Q4_K_M | Q5_K_M | 31K | 79 | no | Check |
| GeForce RTX 3070 | 8 GB | Q4_K_M | Q5_K_M | 31K | 58 | no | Check |
| GeForce RTX 3060 Ti | 8 GB | Q4_K_M | Q5_K_M | 31K | 58 | no | Check |
| GeForce RTX 3060 12 GB | 12 GB | Q6_K | Q8_0 | 97K | 47 | no | Check |
| GeForce RTX 2080 Ti | 11 GB | Q6_K | Q8_0 | 81K | 80 | no | Check |
| GeForce RTX 2060 12 GB | 12 GB | Q6_K | Q8_0 | 97K | 44 | no | Check |
| GeForce GTX 1080 Ti | 11 GB | Q6_K | Q8_0 | 81K | 63 | no | Check |
NVIDIA · GeForce laptop
| Device | Memory | Best quant with headroom | Fits at all | Max context @ Q4_K_M | Tok/s est. | 128K | |
|---|---|---|---|---|---|---|---|
| GeForce RTX 5090 Laptop | 24 GB | Q8_0 | Q8_0 | 125K | 116 | yes | Check |
| GeForce RTX 4090 Laptop | 16 GB | Q8_0 | Q8_0 | 125K | 75 | yes | Check |
| GeForce RTX 4080 Laptop | 12 GB | Q6_K | Q8_0 | 97K | 56 | no | Check |
| GeForce RTX 4070 Laptop | 8 GB | Q4_K_M | Q5_K_M | 31K | 33 | no | Check |
NVIDIA · Workstation
| Device | Memory | Best quant with headroom | Fits at all | Max context @ Q4_K_M | Tok/s est. | 128K | |
|---|---|---|---|---|---|---|---|
| RTX 6000 Ada | 48 GB | Q8_0 | Q8_0 | 125K | 125 | yes | Check |
| RTX 5000 Ada | 32 GB | Q8_0 | Q8_0 | 125K | 75 | yes | Check |
| RTX 4000 Ada | 20 GB | Q8_0 | Q8_0 | 125K | 47 | yes | Check |
| RTX 2000 Ada | 16 GB | Q8_0 | Q8_0 | 125K | 29 | yes | Check |
| RTX A6000 | 48 GB | Q8_0 | Q8_0 | 125K | 100 | yes | Check |
| RTX A5000 | 24 GB | Q8_0 | Q8_0 | 125K | 100 | yes | Check |
| RTX A4000 | 16 GB | Q8_0 | Q8_0 | 125K | 58 | yes | Check |
NVIDIA · Data centre
| Device | Memory | Best quant with headroom | Fits at all | Max context @ Q4_K_M | Tok/s est. | 128K | |
|---|---|---|---|---|---|---|---|
| L40S | 48 GB | Q8_0 | Q8_0 | 125K | 112 | yes | Check |
| L4 | 24 GB | Q8_0 | Q8_0 | 125K | 39 | yes | Check |
| A10 | 24 GB | Q8_0 | Q8_0 | 125K | 78 | yes | Check |
| A100 40 GB | 40 GB | Q8_0 | Q8_0 | 125K | 202 | yes | Check |
| A100 80 GB | 80 GB | Q8_0 | Q8_0 | 125K | 265 | yes | Check |
| H100 PCIe | 80 GB | Q8_0 | Q8_0 | 125K | 260 | yes | Check |
| H100 SXM | 80 GB | Q8_0 | Q8_0 | 125K | 435 | yes | Check |
| H200 SXM | 141 GB | Q8_0 | Q8_0 | 125K | 623 | yes | Check |
| Tesla P40 | 24 GB | Q8_0 | Q8_0 | 125K | 45 | yes | Check |
AMD · Radeon
| Device | Memory | Best quant with headroom | Fits at all | Max context @ Q4_K_M | Tok/s est. | 128K | |
|---|---|---|---|---|---|---|---|
| Radeon RX 9070 XT | 16 GB | Q8_0 | Q8_0 | 125K | 84 | yes | Check |
| Radeon RX 9070 | 16 GB | Q8_0 | Q8_0 | 125K | 84 | yes | Check |
| Radeon RX 7900 XTX | 24 GB | Q8_0 | Q8_0 | 125K | 125 | yes | Check |
| Radeon RX 7900 XT | 20 GB | Q8_0 | Q8_0 | 125K | 104 | yes | Check |
| Radeon RX 7900 GRE | 16 GB | Q8_0 | Q8_0 | 125K | 75 | yes | Check |
| Radeon RX 7800 XT | 16 GB | Q8_0 | Q8_0 | 125K | 81 | yes | Check |
| Radeon RX 7700 XT | 12 GB | Q6_K | Q8_0 | 97K | 56 | no | Check |
| Radeon RX 7600 XT | 16 GB | Q8_0 | Q8_0 | 125K | 37 | yes | Check |
| Radeon RX 6900 XT | 16 GB | Q8_0 | Q8_0 | 125K | 66 | yes | Check |
| Radeon RX 6800 XT | 16 GB | Q8_0 | Q8_0 | 125K | 66 | yes | Check |
| Radeon RX 6700 XT | 12 GB | Q6_K | Q8_0 | 97K | 50 | no | Check |
AMD · Radeon Pro
| Device | Memory | Best quant with headroom | Fits at all | Max context @ Q4_K_M | Tok/s est. | 128K | |
|---|---|---|---|---|---|---|---|
| Radeon Pro W7900 | 48 GB | Q8_0 | Q8_0 | 125K | 112 | yes | Check |
| Radeon Pro W7800 | 32 GB | Q8_0 | Q8_0 | 125K | 75 | yes | Check |
AMD · Instinct
| Device | Memory | Best quant with headroom | Fits at all | Max context @ Q4_K_M | Tok/s est. | 128K | |
|---|---|---|---|---|---|---|---|
| Instinct MI210 | 64 GB | Q8_0 | Q8_0 | 125K | 213 | yes | Check |
| Instinct MI300X | 192 GB | Q8_0 | Q8_0 | 125K | 688 | yes | Check |
AMD · Ryzen AI Max
| Device | Memory | Best quant with headroom | Fits at all | Max context @ Q4_K_M | Tok/s est. | 128K | |
|---|---|---|---|---|---|---|---|
| Ryzen AI Max+ 395 · 128 GB | 128 GB | Q8_0 | Q8_0 | 125K | 31 | yes | Check |
| Ryzen AI Max+ 395 · 64 GB | 64 GB | Q8_0 | Q8_0 | 125K | 31 | yes | Check |
Apple · Apple Silicon
| Device | Memory | Best quant with headroom | Fits at all | Max context @ Q4_K_M | Tok/s est. | 128K | |
|---|---|---|---|---|---|---|---|
| M1 · 16 GB | 16 GB | Q6_K | Q8_0 | 99K | 8.2 | no | Check |
| M1 Pro · 32 GB | 32 GB | Q8_0 | Q8_0 | 125K | 24 | yes | Check |
| M1 Max · 64 GB | 64 GB | Q8_0 | Q8_0 | 125K | 48 | yes | Check |
| M1 Ultra · 128 GB | 128 GB | Q8_0 | Q8_0 | 125K | 96 | yes | Check |
| M2 · 24 GB | 24 GB | Q8_0 | Q8_0 | 125K | 12 | yes | Check |
| M2 Pro · 32 GB | 32 GB | Q8_0 | Q8_0 | 125K | 24 | yes | Check |
| M2 Max · 96 GB | 96 GB | Q8_0 | Q8_0 | 125K | 48 | yes | Check |
| M2 Ultra · 192 GB | 192 GB | Q8_0 | Q8_0 | 125K | 96 | yes | Check |
| M3 · 24 GB | 24 GB | Q8_0 | Q8_0 | 125K | 12 | yes | Check |
| M3 Pro · 36 GB | 36 GB | Q8_0 | Q8_0 | 125K | 18 | yes | Check |
| M3 Max · 64 GB | 64 GB | Q8_0 | Q8_0 | 125K | 48 | yes | Check |
| M3 Max · 128 GB | 128 GB | Q8_0 | Q8_0 | 125K | 48 | yes | Check |
| M3 Ultra · 256 GB | 256 GB | Q8_0 | Q8_0 | 125K | 96 | yes | Check |
| M3 Ultra · 512 GB | 512 GB | Q8_0 | Q8_0 | 125K | 96 | yes | Check |
| M4 · 32 GB | 32 GB | Q8_0 | Q8_0 | 125K | 14 | yes | Check |
| M4 Pro · 48 GB | 48 GB | Q8_0 | Q8_0 | 125K | 33 | yes | Check |
| M4 Pro · 64 GB | 64 GB | Q8_0 | Q8_0 | 125K | 33 | yes | Check |
| M4 Max · 64 GB | 64 GB | Q8_0 | Q8_0 | 125K | 49 | yes | Check |
| M4 Max · 128 GB | 128 GB | Q8_0 | Q8_0 | 125K | 65 | yes | Check |
Intel · Arc
| Device | Memory | Best quant with headroom | Fits at all | Max context @ Q4_K_M | Tok/s est. | 128K | |
|---|---|---|---|---|---|---|---|
| Arc B580 | 12 GB | Q6_K | Q8_0 | 97K | 59 | no | Check |
| Arc B570 | 10 GB | Q6_K | Q6_K | 64K | 49 | no | Check |
| Arc A770 16 GB | 16 GB | Q8_0 | Q8_0 | 125K | 73 | yes | Check |
| Arc A750 | 8 GB | Q4_K_M | Q5_K_M | 31K | 66 | no | Check |
CPU · CPU + system RAM
| Device | Memory | Best quant with headroom | Fits at all | Max context @ Q4_K_M | Tok/s est. | 128K | |
|---|---|---|---|---|---|---|---|
| CPU only · DDR4 dual-channel | 32 GB | Q8_0 | Q8_0 | 125K | 4.6 | yes | Check |
| CPU only · DDR5 dual-channel | 64 GB | Q8_0 | Q8_0 | 125K | 8.1 | yes | Check |
| CPU only · DDR5 8-channel server | 256 GB | Q8_0 | Q8_0 | 125K | 28 | yes | Check |