K2-Horizon MoVA 36B-A4B hardware requirements

Open license · Apache-2.0ChatCodingReasoning

What is K2-Horizon MoVA 36B-A4B?

K2-Horizon MoVA 36B-A4B is the mixture-of-experts member of the K2-Horizon family, which IFM released in September 2026. About 4B of its roughly 37B parameters are active per token, and its attention uses a design the card calls Mixture-of-Values (MoVA). Thinking is always on, with reasoning_effort setting its depth, and the native context is 524,288 tokens. The Apache 2.0 license allows commercial use.

Model card

About the model

Parameters: 37.44B total · 4B active (MoE)
Attention: Full attention
Max context: 524,288 tokens
Released: 2026-09-01
Model card: K2-Horizon MoVA 36B-A4B on Hugging Face

Other sizes in this series

Graphics card memory each model needs at its baseline quant with 8k context, and its verdict on common graphics cards, smallest model first
Model	Parameters	Baseline quant	Memory needed (8k)	GeForce RTX 4060	GeForce RTX 5060 Ti 16GB	GeForce RTX 3090
K2-Horizon 7B	9B	Q4_K_M	6.8 GB	Runs well	Runs great	Runs great
K2-Horizon MoVA 36B-A4Bthis model	37.44B total · 4B active (MoE)	Q4_K_M	24.2 GB	Runs slowly	Runs well	Runs well

GGUF files

Tracked quantizations of K2-Horizon MoVA 36B-A4B, smallest first
Quant	File size	Quality	Published by
Q4_K_Mbaseline	22.37 GB	Good balance	Official GGUF · IFM
Q8_0	39.83 GB	Near-lossless	Official GGUF · IFM

Memory needed by quant and context

Graphics card memory in GB: weights loaded on the GPU + f16 KV cache + compute buffer. Add about 0.6 GB if the GPU also drives your display on Windows. Part of the file stays off the graphics card: llama.cpp keeps the input embedding table in system RAM. A Mac counts the whole file, so it needs a little more.
Quant	4k	8k	16k	32k	64k	128k	256k
Q4_K_M	23.3	24.2	25.9	29.2	36.0	49.5	76.4
Q8_0	40.5	41.3	43.0	46.4	53.1	66.6	93.6

Which GPUs can run K2-Horizon MoVA 36B-A4B?

How to read the verdicts

Runs great: Fully on the GPU at 20 tok/s or more
Runs well: Fully on the GPU at 8–20 tok/s; MoE experts in system RAM at 20 tok/s or more; or at least 90% on the GPU at 8 tok/s or more
Runs slowly: 2–8 tok/s; CPU-only; less than 90% on the GPU; or MoE experts in system RAM below 20 tok/s
Won't run: Does not fit, or under 2 tok/s

Sort

Filter this list by budget, type and context in the calculator

Scroll sideways to see every column.

56 GPUs, Macs and CPU setups at Q4_K_M with 8k context, sorted by verdict and speed
Hardware	Verdict	Speed	Quant	Memory	Runs as	Context	Notes	Price
GeForce RTX 5090	Runs great	est. 149.6 tok/s104.7–194.5theoretical estimate ±30%	Q4_K_M	24.8 / 32.0 GB	Full GPU	up to 32k	—	$4,200 street (as of 2026-08-09)
Apple M5 Ultra · 96 GB	Runs great	est. 82.3 tok/s57.6–106.9theoretical estimate ±30%	Q4_K_M	24.6 / 67.2 GB	Unified memory	up to 64k	—	—
Apple M5 Max (40-core GPU) · 48 GB	Runs great	est. 71.2 tok/s49.9–92.6theoretical estimate ±30%	Q4_K_M	24.6 / 33.6 GB	Unified memory	up to 32k	—	—
Apple M4 Max (40-core GPU) · 48 GB	Runs great	est. 65.1 tok/s45.6–84.7theoretical estimate ±30%	Q4_K_M	24.6 / 33.6 GB	Unified memory	up to 32k	—	—
Apple M3 Ultra · 96 GB	Runs great	est. 63.9 tok/s44.7–83.0theoretical estimate ±30%	Q4_K_M	24.6 / 67.2 GB	Unified memory	up to 32k	—	—
Apple M2 Ultra · 64 GB	Runs great	est. 62.9 tok/s44.0–81.7theoretical estimate ±30%	Q4_K_M	24.6 / 44.8 GB	Unified memory	up to 32k	—	—
Apple M5 Max (32-core GPU) · 36 GB	Runs great	est. 56.9 tok/s39.9–74.0theoretical estimate ±30%	Q4_K_M	24.6 / 25.2 GB	Unified memory	up to 8k	—	—
Apple M4 Max (32-core GPU) · 36 GB	Runs great	est. 51.9 tok/s36.3–67.5theoretical estimate ±30%	Q4_K_M	24.6 / 25.2 GB	Unified memory	up to 8k	—	—
Apple M5 Pro · 48 GB	Runs great	est. 40.7 tok/s28.5–52.9theoretical estimate ±30%	Q4_K_M	24.6 / 33.6 GB	Unified memory	up to 16k	—	—
NVIDIA DGX Spark · 64 GB	Runs great	est. 37.7 tok/s26.4–49.1theoretical estimate ±30%	Q4_K_M	24.6 / 48.2 GB	Unified memory	up to 8k	—	$3,999 launch MSRP (128 GB)
Apple M4 Pro · 48 GB	Runs great	est. 36.8 tok/s25.8–47.8theoretical estimate ±30%	Q4_K_M	24.6 / 33.6 GB	Unified memory	up to 8k	—	—
Ryzen AI Max+ 395 (Strix Halo) · 64 GB	Runs great	est. 35.1 tok/s24.5–45.6theoretical estimate ±30%	Q4_K_M	24.6 / 44.8 GB	Unified memory	up to 8k	—	$1,999 launch MSRP (128 GB)
GeForce RTX 3090 Ti	Runs well	est. 82.6 tok/s57.8–107.4theoretical estimate ±30%	Q4_K_M	23.4 / 24.0 GB + 1.8 GB RAM	MoE experts in RAM	up to 32k	Not fully on the GPU — Runs great needs the whole model in GPU memory	$1,999 launch MSRP
GeForce RTX 4090	Runs well	est. 82.6 tok/s57.8–107.4theoretical estimate ±30%	Q4_K_M	23.4 / 24.0 GB + 1.8 GB RAM	MoE experts in RAM	up to 32k	Not fully on the GPU — Runs great needs the whole model in GPU memory	$2,755 street (as of 2026-08-09)
GeForce RTX 3090	Runs well	est. 79.2 tok/s55.5–103.0theoretical estimate ±30%	Q4_K_M	23.4 / 24.0 GB + 1.8 GB RAM	MoE experts in RAM	up to 32k	Not fully on the GPU — Runs great needs the whole model in GPU memory	$1,050 street (as of 2026-08-09)
GeForce RTX 5090 Laptop	Runs well	est. 77.2 tok/s54.1–100.4theoretical estimate ±30%	Q4_K_M	23.4 / 24.0 GB + 1.8 GB RAM	MoE experts in RAM	up to 32k	Not fully on the GPU — Runs great needs the whole model in GPU memory	—
Radeon RX 7900 XTX	Runs well	est. 68.1 tok/s47.7–88.5theoretical estimate ±30%	Q4_K_M	23.4 / 24.0 GB + 1.8 GB RAM	MoE experts in RAM	up to 32k	Not fully on the GPU — Runs great needs the whole model in GPU memory	$999 launch MSRP
Radeon RX 7900 XT	Runs well	est. 44.5 tok/s31.2–57.9theoretical estimate ±30%	Q4_K_M	19.4 / 20.0 GB + 5.7 GB RAM	MoE experts in RAM	up to 16k	Not fully on the GPU — Runs great needs the whole model in GPU memory	$899 launch MSRP
GeForce RTX 5080	Runs well	est. 39.3 tok/s27.5–51.0theoretical estimate ±30%	Q4_K_M	15.5 / 16.0 GB + 9.6 GB RAM	MoE experts in RAM	up to 16k	Not fully on the GPU — Runs great needs the whole model in GPU memory	$1,256 street (as of 2026-08-09)
GeForce RTX 5070 Ti	Runs well	est. 38.6 tok/s27.0–50.2theoretical estimate ±30%	Q4_K_M	15.5 / 16.0 GB + 9.6 GB RAM	MoE experts in RAM	up to 16k	Not fully on the GPU — Runs great needs the whole model in GPU memory	$949 street (as of 2026-08-09)
GeForce RTX 5080 Laptop	Runs well	est. 38.6 tok/s27.0–50.2theoretical estimate ±30%	Q4_K_M	15.5 / 16.0 GB + 9.6 GB RAM	MoE experts in RAM	up to 16k	Not fully on the GPU — Runs great needs the whole model in GPU memory	—
GeForce RTX 4080 Super	Runs well	est. 36.6 tok/s25.6–47.5theoretical estimate ±30%	Q4_K_M	15.5 / 16.0 GB + 9.6 GB RAM	MoE experts in RAM	up to 8k	Not fully on the GPU — Runs great needs the whole model in GPU memory	$999 launch MSRP
GeForce RTX 4080	Runs well	est. 36.3 tok/s25.4–47.2theoretical estimate ±30%	Q4_K_M	15.5 / 16.0 GB + 9.6 GB RAM	MoE experts in RAM	up to 8k	Not fully on the GPU — Runs great needs the whole model in GPU memory	$1,199 launch MSRP
GeForce RTX 4070 Ti Super	Runs well	est. 35.6 tok/s24.9–46.2theoretical estimate ±30%	Q4_K_M	15.5 / 16.0 GB + 9.6 GB RAM	MoE experts in RAM	up to 8k	Not fully on the GPU — Runs great needs the whole model in GPU memory	$799 launch MSRP
GeForce RTX 4090 Laptop	Runs well	est. 33.8 tok/s23.6–43.9theoretical estimate ±30%	Q4_K_M	15.5 / 16.0 GB + 9.6 GB RAM	MoE experts in RAM	up to 8k	Not fully on the GPU — Runs great needs the whole model in GPU memory	—
Radeon RX 9070	Runs well	est. 32.7 tok/s22.9–42.5theoretical estimate ±30%	Q4_K_M	15.5 / 16.0 GB + 9.6 GB RAM	MoE experts in RAM	up to 8k	Not fully on the GPU — Runs great needs the whole model in GPU memory	$639 street (as of 2026-08-09)
Radeon RX 9070 XT	Runs well	est. 32.7 tok/s22.9–42.5theoretical estimate ±30%	Q4_K_M	15.5 / 16.0 GB + 9.6 GB RAM	MoE experts in RAM	up to 8k	Not fully on the GPU — Runs great needs the whole model in GPU memory	$689 street (as of 2026-08-09)
Radeon RX 7800 XT	Runs well	est. 32.4 tok/s22.7–42.1theoretical estimate ±30%	Q4_K_M	15.5 / 16.0 GB + 9.6 GB RAM	MoE experts in RAM	up to 8k	Not fully on the GPU — Runs great needs the whole model in GPU memory	$499 launch MSRP
GeForce RTX 3080 12GB	Runs well	est. 31.0 tok/s21.7–40.3theoretical estimate ±30%	Q4_K_M	11.6 / 12.0 GB + 13.5 GB RAM	MoE experts in RAM	up to 8k	Not fully on the GPU — Runs great needs the whole model in GPU memory	$799 launch MSRP
GeForce RTX 5060 Ti 16GB	Runs well	est. 30.7 tok/s21.5–39.9theoretical estimate ±30%	Q4_K_M	15.5 / 16.0 GB + 9.6 GB RAM	MoE experts in RAM	up to 8k	Not fully on the GPU — Runs great needs the whole model in GPU memory	$569 street (as of 2026-08-09)
GeForce RTX 5070	Runs slowly	est. 29.0 tok/s20.3–37.7theoretical estimate ±30%	Q4_K_M	11.6 / 12.0 GB + 13.5 GB RAM	MoE experts in RAM	up to 16k	Experts run from system RAM — Runs well needs 30 tok/s or more on this path	$629 street (as of 2026-08-09)
GeForce RTX 5070 Ti Laptop	Runs slowly	est. 29.0 tok/s20.3–37.7theoretical estimate ±30%	Q4_K_M	11.6 / 12.0 GB + 13.5 GB RAM	MoE experts in RAM	up to 16k	Experts run from system RAM — Runs well needs 30 tok/s or more on this path	—
GeForce RTX 2080 Ti	Runs slowly	est. 27.1 tok/s19.0–35.2theoretical estimate ±30%	Q4_K_M	10.6 / 11.0 GB + 14.6 GB RAM	MoE experts in RAM	up to 16k	Experts run from system RAM — Runs well needs 30 tok/s or more on this path	$999 launch MSRP
GeForce RTX 3080 10GB	Runs slowly	est. 27.0 tok/s18.9–35.2theoretical estimate ±30%	Q4_K_M	9.5 / 10.0 GB + 15.7 GB RAM	MoE experts in RAM	up to 8k	Experts run from system RAM — Runs well needs 30 tok/s or more on this path	$699 launch MSRP
GeForce RTX 4070	Runs slowly	est. 26.9 tok/s18.8–35.0theoretical estimate ±30%	Q4_K_M	11.6 / 12.0 GB + 13.5 GB RAM	MoE experts in RAM	up to 16k	Experts run from system RAM — Runs well needs 30 tok/s or more on this path	$599 launch MSRP
GeForce RTX 4070 Super	Runs slowly	est. 26.9 tok/s18.8–35.0theoretical estimate ±30%	Q4_K_M	11.6 / 12.0 GB + 13.5 GB RAM	MoE experts in RAM	up to 16k	Experts run from system RAM — Runs well needs 30 tok/s or more on this path	$599 launch MSRP
GeForce RTX 4080 Laptop	Runs slowly	est. 25.6 tok/s18.0–33.3theoretical estimate ±30%	Q4_K_M	11.6 / 12.0 GB + 13.5 GB RAM	MoE experts in RAM	up to 16k	Experts run from system RAM — Runs well needs 30 tok/s or more on this path	—
Radeon RX 9060 XT 16GB	Runs slowly	est. 25.6 tok/s17.9–33.2theoretical estimate ±30%	Q4_K_M	15.5 / 16.0 GB + 9.6 GB RAM	MoE experts in RAM	up to 32k	Experts run from system RAM — Runs well needs 30 tok/s or more on this path	$449 street (as of 2026-08-09)
GeForce GTX 1080 Ti	Runs slowly	est. 25.5 tok/s17.8–33.1theoretical estimate ±30%	Q4_K_M	10.6 / 11.0 GB + 14.6 GB RAM	MoE experts in RAM	up to 16k	Experts run from system RAM — Runs well needs 30 tok/s or more on this path	$699 launch MSRP
GeForce RTX 4060 Ti 16GB	Runs slowly	est. 25.0 tok/s17.5–32.6theoretical estimate ±30%	Q4_K_M	15.5 / 16.0 GB + 9.6 GB RAM	MoE experts in RAM	up to 32k	Experts run from system RAM — Runs well needs 30 tok/s or more on this path	$499 launch MSRP
GeForce RTX 3060 12GB	Runs slowly	est. 24.1 tok/s16.9–31.3theoretical estimate ±30%	Q4_K_M	11.6 / 12.0 GB + 13.5 GB RAM	MoE experts in RAM	up to 16k	Experts run from system RAM — Runs well needs 30 tok/s or more on this path	$250 street (as of 2026-08-09)
Intel Arc B580	Runs slowly	est. 23.1 tok/s16.1–30.0theoretical estimate ±30%	Q4_K_M	11.6 / 12.0 GB + 13.5 GB RAM	MoE experts in RAM	up to 16k	Experts run from system RAM — Runs well needs 30 tok/s or more on this path	$290 street (as of 2026-08-09)
Apple M6 · 32 GB	Runs slowly	est. 18.3 tok/s12.8–23.8theoretical estimate ±30%	Q4_K_M	24.0 GB RAM	CPU only	up to 16k	CPU inference — capped at Runs slowly	—
Apple M5 · 32 GB	Runs slowly	est. 16.5 tok/s11.6–21.5theoretical estimate ±30%	Q4_K_M	24.0 GB RAM	CPU only	up to 16k	CPU inference — capped at Runs slowly	—
Apple M4 · 32 GB	Runs slowly	est. 12.9 tok/s9.0–16.8theoretical estimate ±30%	Q4_K_M	24.0 GB RAM	CPU only	up to 16k	CPU inference — capped at Runs slowly	—
DDR5-6000 dual-channel CPU · 32 GB	Runs slowly	est. 10.3 tok/s7.2–13.4theoretical estimate ±30%	Q4_K_M	24.0 GB RAM	CPU only	up to 16k	—	—
DDR5-5600 dual-channel CPU · 32 GB	Runs slowly	est. 9.6 tok/s6.7–12.5theoretical estimate ±30%	Q4_K_M	24.0 GB RAM	CPU only	up to 16k	—	—
GeForce RTX 3050 8GB	Runs slowly	est. 9.6 tok/s6.7–12.5theoretical estimate ±30%	Q4_K_M	24.0 GB RAM	CPU only	up to 16k	—	$249 launch MSRP
GeForce RTX 3070	Runs slowly	est. 9.6 tok/s6.7–12.5theoretical estimate ±30%	Q4_K_M	24.0 GB RAM	CPU only	up to 16k	—	$499 launch MSRP
GeForce RTX 4060	Runs slowly	est. 9.6 tok/s6.7–12.5theoretical estimate ±30%	Q4_K_M	24.0 GB RAM	CPU only	up to 16k	—	$299 launch MSRP
GeForce RTX 4060 Laptop	Runs slowly	est. 9.6 tok/s6.7–12.5theoretical estimate ±30%	Q4_K_M	24.0 GB RAM	CPU only	up to 16k	—	—
GeForce RTX 4060 Ti 8GB	Runs slowly	est. 9.6 tok/s6.7–12.5theoretical estimate ±30%	Q4_K_M	24.0 GB RAM	CPU only	up to 16k	—	$399 launch MSRP
GeForce RTX 4070 Laptop	Runs slowly	est. 9.6 tok/s6.7–12.5theoretical estimate ±30%	Q4_K_M	24.0 GB RAM	CPU only	up to 16k	—	—
GeForce RTX 5060	Runs slowly	est. 9.6 tok/s6.7–12.5theoretical estimate ±30%	Q4_K_M	24.0 GB RAM	CPU only	up to 16k	—	$339 street (as of 2026-08-09)
GeForce RTX 5060 Ti 8GB	Runs slowly	est. 9.6 tok/s6.7–12.5theoretical estimate ±30%	Q4_K_M	24.0 GB RAM	CPU only	up to 16k	—	$429 street (as of 2026-08-09)
DDR4-3200 dual-channel CPU · 32 GB	Runs slowly	est. 5.5 tok/s3.9–7.2theoretical estimate ±30%	Q4_K_M	24.0 GB RAM	CPU only	up to 16k	—	—

Models with a thinking mode spend extra tokens thinking, so their speed thresholds are 1.5× stricter (30 / 12 / 3 tok/s). With thinking turned off, a model may do one grade better.

Cheapest GPUs that run it

Runs greatGeForce RTX 5090$4,200 street (as of 2026-08-09), est. 149.6 tok/s (104.7–194.5, theoretical estimate ±30%).
Runs wellGeForce RTX 5060 Ti 16GB$569 street (as of 2026-08-09), est. 30.7 tok/s (21.5–39.9, theoretical estimate ±30%).

Measured results for K2-Horizon MoVA 36B-A4B

No public measurements for this model yet.

Frequently asked questions

How much VRAM does K2-Horizon MoVA 36B-A4B need?

At Q4_K_M the file is 22.37 GB; with 8k context a graphics card needs about 24.2 GB (weights 21.99 GB, KV cache 1.6 GB, compute buffer 0.6 GB). Part of the file stays off the graphics card: llama.cpp keeps the input embedding table in system RAM. A Mac counts the whole file, so it needs a little more.

What is the cheapest GPU that runs K2-Horizon MoVA 36B-A4B well?

Cheapest that runs it great: GeForce RTX 5090 at $4,200 (street price as of 2026-08-09), est. 149.6 tok/s (104.7–194.5, theoretical estimate ±30%). Cheaper and still Runs well: GeForce RTX 5060 Ti 16GB at $569 (street price as of 2026-08-09), est. 30.7 tok/s (21.5–39.9, theoretical estimate ±30%).

Can K2-Horizon MoVA 36B-A4B run on an 8 GB, 16 GB or 24 GB GPU?

At Q4_K_M with 8k context: GeForce RTX 4060 — Runs slowly, est. 9.6 tok/s (6.7–12.5, theoretical estimate ±30%); GeForce RTX 5060 Ti 16GB — Runs well, est. 30.7 tok/s (21.5–39.9, theoretical estimate ±30%); GeForce RTX 3090 — Runs well, est. 79.2 tok/s (55.5–103.0, theoretical estimate ±30%).

Speeds are estimates from memory bandwidth, calibrated against public benchmarks, and each one comes with an error band and a confidence label. Real results vary with drivers, backend, context length and thermals.