Kolibri 1 (78B-A3.5B) on Intel Arc A770 16GB

Can the Intel Arc A770 16GB run Kolibri 1 (78B-A3.5B)?

Yes, but slowly (heavy RAM offload)

Slow (heavy RAM offload). At Q2_K it needs about 32.8 GB of 16.0 GB usable, with an estimated speed of 53.1 tok/s.

Best quant: Q2_K
Generation speed: 53.1 tok/s
Prompt processing: 3,894 tok/s
First token (1K prompt): 0.3 s
Largest safe context: —

Best app to run it on the Intel Arc A770 16GB

On Windows

llama.cpp

Command line + local API · runs on llama.cpp

Splits the model between GPU memory and system RAM, which this setup needs.
Gives you direct control of how many layers sit on the GPU, the best way to squeeze a model that doesn’t quite fit.
Supports Intel Arc through SYCL and Vulkan.

llama-server -hf bartowski/Kolibri-1-GGUF:Q2_K

LM Studio Splits the model between GPU memory and system RAM, which this setup needs.
Jan Splits the model between GPU memory and system RAM, which this setup needs.

On Linux

llama.cpp

Command line + local API · runs on llama.cpp

Splits the model between GPU memory and system RAM, which this setup needs.
Gives you direct control of how many layers sit on the GPU, the best way to squeeze a model that doesn’t quite fit.
Supports Intel Arc through SYCL and Vulkan.

llama-server -hf bartowski/Kolibri-1-GGUF:Q2_K

LM Studio Splits the model between GPU memory and system RAM, which this setup needs.
Jan Splits the model between GPU memory and system RAM, which this setup needs.

Want an agent? Kolibri 1 (78B-A3.5B) supports tool use, so you can pair a llama.cpp or MLX LM server with one of these. Apps that run on this computer come first.

Coding agents

Personal assistants

How we choose: the files published for this model, your operating system and GPU, whether it fits fully in memory, and published speed tests (MLX vs llama.cpp on Apple silicon, vLLM vs llama.cpp vs Ollama on NVIDIA). Details.

Download Kolibri 1 (78B-A3.5B)

Recommended for the Intel Arc A770 16GB: Q2_K. Every other quantization has its own download link in the table below.

Quantized files · GGUF

Works in LM Studio, llama.cpp, Jan and Ollama. No sign-in needed.

Download the Q2_K file (opens in a new tab)

huggingface.co/bartowski/Kolibri-1-GGUF (opens in a new tab)

ollama run hf.co/bartowski/Kolibri-1-GGUF:Q2_K

Original weights · Hugging Face

Full-precision files from Aleph Alpha, for fine-tuning, vLLM, Transformers.

huggingface.co/Aleph-Alpha/Kolibri-1-BF16 (opens in a new tab)

No sign-in needed.

Official repo · GitHub

Reference code, inference scripts and docs from Aleph Alpha. The weights themselves download from Hugging Face.

github.com/Aleph-Alpha/aleph-alpha-inference (opens in a new tab)

Aleph Alpha's inference package with the vLLM plugin needed to serve Kolibri in vLLM.

git clone https://github.com/Aleph-Alpha/aleph-alpha-inference

Links checked 1 day ago against Hugging Face, Ollama and GitHub.

How it’s rated

Quality 63Good quality (B) · 63 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking.: How capable the model itself is at Q2_K, from its editorial quality tier scaled by the quantization or precision. It’s the same on any hardware that runs this variant. Scale: A+ 85+ · A 70+ · B 55+ · C 40+ · D under 40.
Runs Slowheavy RAM offload: How well it runs on the Intel Arc A770 16GB: speed, capped by how well it fits in memory. Runs great: 30+ tokens/s (or near the image, video or decision speed target) · Runs well: 10+ tokens/s · Slow: 3+ tokens/s · Very slow: under 3 · Won’t run: doesn’t fit even with offload. A tight fit or RAM offload caps it at Runs well; heavy offload caps it at Slow.

Match score

To order lists, each result also gets a match score out of 100 from four parts. It favors the most capable model that still runs well.

Model qualityEditorial quality tier of the model, scaled by the quantization or precision used.	28.7 / 45
Memory fitFull marks when everything sits in GPU memory; fewer when part spills to system RAM.	4.0 / 25
SpeedGeneration speed against a 40 tokens/s target for LLMs, or render time against a target for image and video.	15.0 / 15
Memory useRewards using the memory you have, so a bigger, better variant beats a tiny one that leaves most memory idle.	15.0 / 15
Match score	62.7 / 100

The match score orders results; it is not shown as a grade. Quality and Runs are the two ratings in the tables.

Every quantization on the Intel Arc A770 16GB

Quant	Memory needed	Available	Speed	Runs	Quality	Download
`Q2_K`	32.8 GB	16.0 GB	53.1 tok/s	Slowheavy RAM offload	63Good quality (B) · 63 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking.	GGUF (opens in a new tab)
`Q3_K_M`	39.8 GB	16.0 GB	—	Won't run	76Great quality (A) · 76 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 10/100 (quality, fit, speed and memory use) sets the ranking.	GGUF (opens in a new tab)
`Q4_K_M`	49.2 GB	16.0 GB	—	Won't run	86Excellent quality (A+) · 86 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 11/100 (quality, fit, speed and memory use) sets the ranking.	GGUF (opens in a new tab)
`Q5_K_M`	57.0 GB	16.0 GB	—	Won't run	88Excellent quality (A+) · 88 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 11/100 (quality, fit, speed and memory use) sets the ranking.	GGUF · folder (opens in a new tab)
`Q6_K`	65.6 GB	16.0 GB	—	Won't run	90Excellent quality (A+) · 90 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 12/100 (quality, fit, speed and memory use) sets the ranking.	GGUF · folder (opens in a new tab)
`Q8_0`	84.4 GB	16.0 GB	—	Won't run	91Excellent quality (A+) · 91 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 12/100 (quality, fit, speed and memory use) sets the ranking.	GGUF · folder (opens in a new tab)
`F16`	158 GB	16.0 GB	—	Won't run	91Excellent quality (A+) · 91 of 100: how capable the model is at F16, whatever it runs on. Match score 12/100 (quality, fit, speed and memory use) sets the ranking.	GGUF · folder (opens in a new tab) Original weights (opens in a new tab)

Memory figures assume 8K of context. Longer conversations or documents grow the KV cache.

Other hardware that runs it

91Excellent quality (A+) · 91 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 95/100 (quality, fit, speed and memory use) sets the ranking. Apple M2 Ultra (192 GB) Q8_0 · 84.4 tok/s · Runs great
91Excellent quality (A+) · 91 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 95/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA GB10 (128 GB) Q8_0 · 41.5 tok/s · Runs great
90Excellent quality (A+) · 90 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 95/100 (quality, fit, speed and memory use) sets the ranking. AMD Ryzen AI Max+ 395 (128 GB) Q6_K · 45.4 tok/s · Runs great
90Excellent quality (A+) · 90 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 95/100 (quality, fit, speed and memory use) sets the ranking. Apple M4 Max (128 GB) Q6_K · 81.6 tok/s · Runs great
90Excellent quality (A+) · 90 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 95/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA RTX PRO 6000 Blackwell Q6_K · 170 tok/s · Runs great
90Excellent quality (A+) · 90 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 95/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA RTX Spark N1X (6144-core GPU) (128 GB) Q6_K · 58.3 tok/s · Runs great

Estimates come from published hardware specs and our open formulas, not from running the model on your machine. Real speed varies with runtime, drivers and settings. Numbers marked “Measured” come from a linked benchmark.