Kolibri 1 hardware requirements

Open license · Apache-2.0ChatCodingReasoning

What is Kolibri 1?

Kolibri 1 is a mixture-of-experts reasoning model that Aleph Alpha released in October 2026, focused on German and English. About 3.46B of its 78.1B parameters are active per token, and it takes text input only. Its model card lists multi-step reasoning, retrieval-augmented generation, agentic tool calling and coding. Thinking is on by default and can be switched off with enable_thinking=false; the native context is 262,144 tokens, and the Apache 2.0 license allows commercial use.

Model card · Release post

About the model

Parameters: 78.1B total · 3.46B active (MoE)
Attention: Sliding-window attention
Max context: 262,144 tokens
Released: 2026-10-02
Model card: Kolibri 1 on Hugging Face

GGUF files

Tracked quantizations of Kolibri 1, smallest first
Quant	File size	Quality	Published by
Q2_K	29.92 GB	Heavy loss	Community GGUF · bartowski
IQ4_XS	43.63 GB	Good balance	Community GGUF · bartowski
Q4_K_Mbaseline	49.76 GB	Good balance	Community GGUF · bartowski
Q8_0	83.14 GB	Near-lossless	Community GGUF · bartowski

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
Q2_K	30.2	30.3	30.5	31.0	32.0	33.9	37.8
IQ4_XS	43.9	44.0	44.3	44.8	45.7	47.7	51.5
Q4_K_M	50.0	50.2	50.4	50.9	51.9	53.8	57.7
Q8_0	83.4	83.5	83.8	84.3	85.2	87.2	91.1

Which GPUs can run Kolibri 1?

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
Apple M5 Ultra · 96 GB	Runs great	est. 108.1 tok/s86.5–129.8calibrated estimate ±20%	Q4_K_M	50.5 / 67.2 GB	Unified memory	up to 256k	—	—
Apple M5 Max (40-core GPU) · 128 GB	Runs great	est. 96.4 tok/s77.1–115.7calibrated estimate ±20%	Q4_K_M	50.5 / 89.6 GB	Unified memory	up to 256k	—	—
Apple M4 Max (40-core GPU) · 128 GB	Runs great	est. 88.5 tok/s70.8–106.2calibrated estimate ±20%	Q4_K_M	50.5 / 89.6 GB	Unified memory	up to 256k	—	—
Apple M3 Ultra · 96 GB	Runs great	est. 85.3 tok/s68.2–102.4calibrated estimate ±20%	Q4_K_M	50.5 / 67.2 GB	Unified memory	up to 256k	—	—
Apple M2 Ultra · 128 GB	Runs great	est. 84.0 tok/s67.2–100.8calibrated estimate ±20%	Q4_K_M	50.5 / 89.6 GB	Unified memory	up to 256k	—	—
NVIDIA DGX Spark · 128 GB	Runs great	est. 52.2 tok/s41.7–62.6calibrated estimate ±20%	Q4_K_M	50.5 / 116.9 GB	Unified memory	up to 128k	—	$3,999 launch MSRP (128 GB)
Ryzen AI Max+ 395 (Strix Halo) · 128 GB	Runs great	est. 48.0 tok/s38.4–57.6calibrated estimate ±20%	Q4_K_M	50.5 / 89.6 GB	Unified memory	up to 128k	—	$1,999 launch MSRP (128 GB)
GeForce RTX 5090	Runs well	est. 83.3 tok/s58.3–108.3theoretical estimate ±30%	Q4_K_M	30.1 / 32.0 GB + 21.0 GB RAM	MoE experts in RAM	up to 128k	Not fully on the GPU — Runs great needs the whole model in GPU memory Q2_K (heavy quality loss): Runs great	$4,200 street (as of 2026-08-09)
Apple M5 Pro · 64 GB	Runs slowly	est. 55.6 tok/s39.0–72.3theoretical estimate ±30%	Q4_K_M	49.9 GB RAM	CPU only	up to 256k	CPU inference — capped at Runs slowly Try IQ4_XS: Runs great	—
Apple M4 Pro · 64 GB	Runs slowly	est. 49.5 tok/s34.6–64.3theoretical estimate ±30%	Q4_K_M	49.9 GB RAM	CPU only	up to 256k	CPU inference — capped at Runs slowly Try IQ4_XS: Runs great	—
DDR5-6000 dual-channel CPU · 64 GB	Runs slowly	est. 17.4 tok/s13.9–20.9calibrated estimate ±20%	Q4_K_M	49.9 GB RAM	CPU only	up to 256k	CPU inference — capped at Runs slowly	—
DDR5-5600 dual-channel CPU · 64 GB	Runs slowly	est. 16.2 tok/s13.0–19.5calibrated estimate ±20%	Q4_K_M	49.9 GB RAM	CPU only	up to 256k	CPU inference — capped at Runs slowly	—
DDR4-3200 dual-channel CPU · 64 GB	Runs slowly	est. 9.3 tok/s7.4–11.1calibrated estimate ±20%	Q4_K_M	49.9 GB RAM	CPU only	up to 128k	—	—
Apple M5 Max (32-core GPU) · 36 GB	Won't runTry Q2_K (heavy quality loss): Runs slowly	est. 92.9 tok/s65.0–120.7theoretical estimate ±30%	Q2_K	30.1 GB RAM	CPU only	up to 64k	CPU inference — capped at Runs slowly	—
GeForce RTX 5080	Won't runTry Q2_K (heavy quality loss): Runs well	est. 86.5 tok/s60.6–112.5theoretical estimate ±30%	Q2_K	15.0 / 16.0 GB + 16.3 GB RAM	MoE experts in RAM	up to 256k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	$1,256 street (as of 2026-08-09)
Apple M4 Max (32-core GPU) · 36 GB	Won't runTry Q2_K (heavy quality loss): Runs slowly	est. 85.3 tok/s59.7–110.9theoretical estimate ±30%	Q2_K	30.1 GB RAM	CPU only	up to 64k	CPU inference — capped at Runs slowly	—
GeForce RTX 5070 Ti	Won't runTry Q2_K (heavy quality loss): Runs well	est. 84.6 tok/s59.2–110.0theoretical estimate ±30%	Q2_K	15.0 / 16.0 GB + 16.3 GB RAM	MoE experts in RAM	up to 256k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	$949 street (as of 2026-08-09)
GeForce RTX 5080 Laptop	Won't runTry Q2_K (heavy quality loss): Runs well	est. 84.6 tok/s59.2–110.0theoretical estimate ±30%	Q2_K	15.0 / 16.0 GB + 16.3 GB RAM	MoE experts in RAM	up to 256k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	—
GeForce RTX 4080 Super	Won't runTry Q2_K (heavy quality loss): Runs well	est. 78.8 tok/s55.2–102.5theoretical estimate ±30%	Q2_K	15.0 / 16.0 GB + 16.3 GB RAM	MoE experts in RAM	up to 256k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	$999 launch MSRP
GeForce RTX 4080	Won't runTry Q2_K (heavy quality loss): Runs well	est. 78.0 tok/s54.6–101.5theoretical estimate ±30%	Q2_K	15.0 / 16.0 GB + 16.3 GB RAM	MoE experts in RAM	up to 256k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	$1,199 launch MSRP
GeForce RTX 3080 12GB	Won't runTry Q2_K (heavy quality loss): Runs well	est. 76.9 tok/s53.9–100.0theoretical estimate ±30%	Q2_K	11.2 / 12.0 GB + 20.1 GB RAM	MoE experts in RAM	up to 256k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	$799 launch MSRP
GeForce RTX 4070 Ti Super	Won't runTry Q2_K (heavy quality loss): Runs well	est. 76.1 tok/s53.3–98.9theoretical estimate ±30%	Q2_K	15.0 / 16.0 GB + 16.3 GB RAM	MoE experts in RAM	up to 256k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	$799 launch MSRP
GeForce RTX 4090 Laptop	Won't runTry Q2_K (heavy quality loss): Runs well	est. 71.3 tok/s49.9–92.7theoretical estimate ±30%	Q2_K	15.0 / 16.0 GB + 16.3 GB RAM	MoE experts in RAM	up to 256k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	—
GeForce RTX 3090 Ti	Won't runTry IQ4_XS: Runs well	est. 70.1 tok/s49.0–91.1theoretical estimate ±30%	IQ4_XS	22.6 / 24.0 GB + 22.3 GB RAM	MoE experts in RAM	up to 128k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	$1,999 launch MSRP
GeForce RTX 4090	Won't runTry IQ4_XS: Runs well	est. 70.1 tok/s49.0–91.1theoretical estimate ±30%	IQ4_XS	22.6 / 24.0 GB + 22.3 GB RAM	MoE experts in RAM	up to 128k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	$2,755 street (as of 2026-08-09)
GeForce RTX 5070	Won't runTry Q2_K (heavy quality loss): Runs well	est. 69.7 tok/s48.8–90.6theoretical estimate ±30%	Q2_K	11.2 / 12.0 GB + 20.1 GB RAM	MoE experts in RAM	up to 256k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	$629 street (as of 2026-08-09)
GeForce RTX 5070 Ti Laptop	Won't runTry Q2_K (heavy quality loss): Runs well	est. 69.7 tok/s48.8–90.6theoretical estimate ±30%	Q2_K	11.2 / 12.0 GB + 20.1 GB RAM	MoE experts in RAM	up to 256k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	—
GeForce RTX 3080 10GB	Won't runTry Q2_K (heavy quality loss): Runs well	est. 69.2 tok/s48.4–89.9theoretical estimate ±30%	Q2_K	9.0 / 10.0 GB + 22.3 GB RAM	MoE experts in RAM	up to 128k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	$699 launch MSRP
GeForce RTX 3090	Won't runTry IQ4_XS: Runs well	est. 68.5 tok/s47.9–89.0theoretical estimate ±30%	IQ4_XS	22.6 / 24.0 GB + 22.3 GB RAM	MoE experts in RAM	up to 128k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	$1,050 street (as of 2026-08-09)
GeForce RTX 5090 Laptop	Won't runTry IQ4_XS: Runs well	est. 67.5 tok/s47.3–87.8theoretical estimate ±30%	IQ4_XS	22.6 / 24.0 GB + 22.3 GB RAM	MoE experts in RAM	up to 128k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	—
GeForce RTX 2080 Ti	Won't runTry Q2_K (heavy quality loss): Runs well	est. 65.9 tok/s46.1–85.7theoretical estimate ±30%	Q2_K	10.1 / 11.0 GB + 21.2 GB RAM	MoE experts in RAM	up to 128k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	$999 launch MSRP
Radeon RX 9070	Won't runTry Q2_K (heavy quality loss): Runs well	est. 65.5 tok/s45.8–85.1theoretical estimate ±30%	Q2_K	15.0 / 16.0 GB + 16.3 GB RAM	MoE experts in RAM	up to 256k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	$639 street (as of 2026-08-09)
Radeon RX 9070 XT	Won't runTry Q2_K (heavy quality loss): Runs well	est. 65.5 tok/s45.8–85.1theoretical estimate ±30%	Q2_K	15.0 / 16.0 GB + 16.3 GB RAM	MoE experts in RAM	up to 256k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	$689 street (as of 2026-08-09)
Radeon RX 7800 XT	Won't runTry Q2_K (heavy quality loss): Runs well	est. 64.8 tok/s45.3–84.2theoretical estimate ±30%	Q2_K	15.0 / 16.0 GB + 16.3 GB RAM	MoE experts in RAM	up to 256k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	$499 launch MSRP
GeForce RTX 5060 Ti 16GB	Won't runTry Q2_K (heavy quality loss): Runs well	est. 63.3 tok/s44.3–82.3theoretical estimate ±30%	Q2_K	15.0 / 16.0 GB + 16.3 GB RAM	MoE experts in RAM	up to 128k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	$569 street (as of 2026-08-09)
GeForce RTX 4070	Won't runTry Q2_K (heavy quality loss): Runs well	est. 62.2 tok/s43.5–80.9theoretical estimate ±30%	Q2_K	11.2 / 12.0 GB + 20.1 GB RAM	MoE experts in RAM	up to 256k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	$599 launch MSRP
GeForce RTX 4070 Super	Won't runTry Q2_K (heavy quality loss): Runs well	est. 62.2 tok/s43.5–80.9theoretical estimate ±30%	Q2_K	11.2 / 12.0 GB + 20.1 GB RAM	MoE experts in RAM	up to 256k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	$599 launch MSRP
Radeon RX 7900 XTX	Won't runTry IQ4_XS: Runs well	est. 60.7 tok/s42.5–78.9theoretical estimate ±30%	IQ4_XS	22.6 / 24.0 GB + 22.3 GB RAM	MoE experts in RAM	up to 128k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	$999 launch MSRP
GeForce GTX 1080 Ti	Won't runTry Q2_K (heavy quality loss): Runs well	est. 59.9 tok/s41.9–77.9theoretical estimate ±30%	Q2_K	10.1 / 11.0 GB + 21.2 GB RAM	MoE experts in RAM	up to 128k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	$699 launch MSRP
GeForce RTX 4080 Laptop	Won't runTry Q2_K (heavy quality loss): Runs well	est. 58.1 tok/s40.6–75.5theoretical estimate ±30%	Q2_K	11.2 / 12.0 GB + 20.1 GB RAM	MoE experts in RAM	up to 128k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	—
GeForce RTX 3070	Won't runTry Q2_K (heavy quality loss): Runs well	est. 55.3 tok/s38.7–71.9theoretical estimate ±30%	Q2_K	7.3 / 8.0 GB + 23.9 GB RAM	MoE experts in RAM	up to 128k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	$499 launch MSRP
GeForce RTX 5060	Won't runTry Q2_K (heavy quality loss): Runs well	est. 55.3 tok/s38.7–71.9theoretical estimate ±30%	Q2_K	7.3 / 8.0 GB + 23.9 GB RAM	MoE experts in RAM	up to 128k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	$339 street (as of 2026-08-09)
GeForce RTX 5060 Ti 8GB	Won't runTry Q2_K (heavy quality loss): Runs well	est. 55.3 tok/s38.7–71.9theoretical estimate ±30%	Q2_K	7.3 / 8.0 GB + 23.9 GB RAM	MoE experts in RAM	up to 128k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	$429 street (as of 2026-08-09)
Radeon RX 7900 XT	Won't runTry IQ4_XS: Runs well	est. 53.7 tok/s37.6–69.8theoretical estimate ±30%	IQ4_XS	18.6 / 20.0 GB + 26.4 GB RAM	MoE experts in RAM	up to 32k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	$899 launch MSRP
GeForce RTX 3060 12GB	Won't runTry Q2_K (heavy quality loss): Runs well	est. 53.1 tok/s37.2–69.0theoretical estimate ±30%	Q2_K	11.2 / 12.0 GB + 20.1 GB RAM	MoE experts in RAM	up to 128k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	$250 street (as of 2026-08-09)
GeForce RTX 4060 Ti 16GB	Won't runTry Q2_K (heavy quality loss): Runs well	est. 49.5 tok/s34.7–64.4theoretical estimate ±30%	Q2_K	15.0 / 16.0 GB + 16.3 GB RAM	MoE experts in RAM	up to 64k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	$499 launch MSRP
Radeon RX 9060 XT 16GB	Won't runTry Q2_K (heavy quality loss): Runs well	est. 49.1 tok/s34.4–63.8theoretical estimate ±30%	Q2_K	15.0 / 16.0 GB + 16.3 GB RAM	MoE experts in RAM	up to 128k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	$449 street (as of 2026-08-09)
Intel Arc B580	Won't runTry Q2_K (heavy quality loss): Runs well	est. 48.6 tok/s34.0–63.2theoretical estimate ±30%	Q2_K	11.2 / 12.0 GB + 20.1 GB RAM	MoE experts in RAM	up to 128k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	$290 street (as of 2026-08-09)
GeForce RTX 4060 Ti 8GB	Won't runTry Q2_K (heavy quality loss): Runs well	est. 44.9 tok/s31.4–58.4theoretical estimate ±30%	Q2_K	7.3 / 8.0 GB + 23.9 GB RAM	MoE experts in RAM	up to 64k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	$399 launch MSRP
GeForce RTX 4060	Won't runTry Q2_K (heavy quality loss): Runs well	est. 43.5 tok/s30.5–56.6theoretical estimate ±30%	Q2_K	7.3 / 8.0 GB + 23.9 GB RAM	MoE experts in RAM	up to 64k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	$299 launch MSRP
GeForce RTX 4060 Laptop	Won't runTry Q2_K (heavy quality loss): Runs well	est. 42.1 tok/s29.5–54.7theoretical estimate ±30%	Q2_K	7.3 / 8.0 GB + 23.9 GB RAM	MoE experts in RAM	up to 64k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	—
GeForce RTX 4070 Laptop	Won't runTry Q2_K (heavy quality loss): Runs well	est. 42.1 tok/s29.5–54.7theoretical estimate ±30%	Q2_K	7.3 / 8.0 GB + 23.9 GB RAM	MoE experts in RAM	up to 64k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs well	—
GeForce RTX 3050 8GB	Won't runTry Q2_K (heavy quality loss): Runs well	est. 39.0 tok/s27.3–50.7theoretical estimate ±30%	Q2_K	7.3 / 8.0 GB + 23.9 GB RAM	MoE experts in RAM	up to 32k	Not fully on the GPU — Runs great needs the whole model in GPU memory With 64 GB RAM: Runs slowly	$249 launch MSRP
Apple M4 · 32 GB	Won't run	—	Q4_K_M	needs 50.5 GB	—	—	Needs about 50.5 GB; 28.0 GB of memory is free (22.4 GB usable by the GPU)	—
Apple M5 · 32 GB	Won't run	—	Q4_K_M	needs 50.5 GB	—	—	Needs about 50.5 GB; 28.0 GB of memory is free (22.4 GB usable by the GPU)	—
Apple M6 · 32 GB	Won't run	—	Q4_K_M	needs 50.5 GB	—	—	Needs about 50.5 GB; 28.0 GB of memory is free (22.4 GB usable by the GPU)	—

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 wellGeForce RTX 5090$4,200 street (as of 2026-08-09), est. 83.3 tok/s (58.3–108.3, theoretical estimate ±30%).

Measured results for Kolibri 1

No public measurements for this model yet.

Frequently asked questions

How much VRAM does Kolibri 1 need?

At Q4_K_M the file is 49.76 GB; with 8k context a graphics card needs about 50.2 GB (weights 49.41 GB, KV cache 0.2 GB, compute buffer 0.6 GB). The smallest tracked file (Q2_K (heavy quality loss)) needs about 30.3 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 Kolibri 1 well?

No priced GPU runs it great; the cheapest that runs it well is GeForce RTX 5090 at $4,200 (street price as of 2026-08-09), est. 83.3 tok/s (58.3–108.3, theoretical estimate ±30%).

Can Kolibri 1 run on an 8 GB, 16 GB or 24 GB GPU?

At Q4_K_M with 8k context: GeForce RTX 4060 — Won't run; GeForce RTX 5060 Ti 16GB — Won't run; GeForce RTX 3090 — Won't run.

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.