yandexAliceAIForCausalLM

AliceAI Foundation 80B A3B Base — Hardware Requirements & GPU Compatibility

Chat

AliceAI-Foundation-80B-A3B-Base is a base language model from Yandex, built on a hybrid attention and Mixture-of-Experts architecture and trained entirely from scratch. It has 80 billion total parameters with about 3 billion active per token, routed across 512 experts. As a pretrained checkpoint rather than a chat model, it is intended for fine-tuning or research use; its publisher reports results on par with larger open models in math, coding, and reasoning, with particular strength in Russian-language factual knowledge. It supports a 262,144-token context window and is released under the Apache 2.0 license. At roughly 80 billion total parameters, 4-bit quantization needs around 46GB of memory, putting it beyond a single consumer GPU and into multi-GPU or workstation territory.

81.3B params · MoE5.5B active262K contextApache 2.0Sep 20263.1K downloads409 likes13.2K quant downloads

Your hardware

Verdicts below are for the reference RTX 4090. Set yours to personalise the page.

Detection reads your browser's GPU name with no data sent anywhere.

AliceAI Foundation 80B A3B Base runs on a 64 GB+ Mac or a workstation GPU

Best local pick → Mac Pro M2 Ultra (64 GB) · Q4_K_M · 49.9 GB · ~64 tok/s

Macs from 48 GB (Q3_K_L · ~62 tok/s)

Apache 2.0

Not set up for it?

Reference GPU · RTX 4090 24 GB

Estimates, ±20% · how we estimate

Run AliceAI Foundation 80B A3B Base

One recommended way. Every other method, OS and quant is in the step-by-step guide.Each way opens a step-by-step guide for your OS and quant.

Ways to runOther ways

each opens step-by-step

Run in cloud

Runs on A100 80GB (30 GB headroom) · Q4_K_M

Generation speed: ~194 tok/s; generation speed
Cost per 1M output tokens: $1.62; per 1M output tokens

Deployon Vast.aiBest valuefrom $1.14/hr

Compare GPUs →

Run AliceAI Foundation 80B A3B Base

49.9 GB VRAM at 16K context

Works on macOS, Windows and Linux.

Install Ollama (Linux)
$curl -fsSL https://ollama.com/install.sh | sh
Install Ollama (macOS / Windows)
Download and run the app; it installs the ollama command.
Download Ollama ↗
Run Q4_K_M
$ollama run hf.co/Yamada114514/AliceAI-Foundation-80B-A3B-Base-GGUF:Q4_K_M
Raise the context window
Type it inside the chat. For the API server instead, start it with OLLAMA_CONTEXT_LENGTH=16384 ollama serve.
$/set parameter num_ctx 16384
A larger context needs more memory than the weights alone.
Use the API
Ollama serves its API on http://localhost:11434.

Works on macOS, Windows and Linux.

Install LM Studio
Desktop app for macOS (Apple silicon), Windows and Linux.
lmstudio.ai ↗
Download the model
In the app press Ctrl/Cmd + Shift + M and paste https://huggingface.co/Yamada114514/AliceAI-Foundation-80B-A3B-Base-GGUF, then pick Q4_K_M. Or use the CLI (run LM Studio once first):
$lms get https://huggingface.co/Yamada114514/AliceAI-Foundation-80B-A3B-Base-GGUF@Q4_K_M
Open on Hugging Face ↗
Load and chat
Open the model loader, select the model and set Context Length to 16384. Leave GPU offload at maximum if it fits.
Start the local server (optional)
$lms server start

Works on macOS, Windows and Linux.

Install llama.cpp (macOS)
MacPorts, Nix and conda-forge also work.
$brew install llama.cpp
Install options ↗
Install llama.cpp (Windows)
$winget install llama.cpp
Install llama.cpp (Linux)
Download the prebuilt llama-*-bin-ubuntu-* archive for your GPU backend (CPU, Vulkan) from the releases page, unpack it and run the binaries from that folder. Homebrew (brew install llama.cpp) and conda-forge also work on Linux.
llama.cpp releases ↗
Start the server
Downloads the model on first run. The chat UI and an OpenAI-compatible API are at http://localhost:8080.
$llama-server -hf Yamada114514/AliceAI-Foundation-80B-A3B-Base-GGUF:Q4_K_M -c 16384 -ngl 99 --port 8080

Works on macOS, Windows and Linux.

Option A: Docker Model Runner
Docker Desktop: Settings > AI > Enable Docker Model Runner. Docker Engine on Linux: install the docker-model-plugin package.
$docker model run hf.co/Yamada114514/AliceAI-Foundation-80B-A3B-Base-GGUF:Q4_K_M
Model Runner docs ↗
The quant tag after the repo is how Hugging Face cards write it; if your Docker version rejects it, drop the tag and pick the quant by repo.
Option B: llama.cpp server image (NVIDIA)
Needs the NVIDIA Container Toolkit. Models are cached in the llama-cache volume. Open http://localhost:8080.
$docker run --gpus all -p 8080:8080 -v llama-cache:/root/.cache ghcr.io/ggml-org/llama.cpp:server-cuda -hf Yamada114514/AliceAI-Foundation-80B-A3B-Base-GGUF:Q4_K_M -c 16384 -ngl 99 --host 0.0.0.0 --port 8080
Image tags (cuda13, vulkan, rocm) ↗
Cache path inside the image is not in the docs; remove the -v flag if the volume complains.

Linux only.

Install vLLM
Linux only; on Windows run it inside WSL2. NVIDIA GPU shown; AMD ROCm, Intel and CPU builds have their own install pages.
$uv venv --python 3.12 --seed
$source .venv/bin/activate
$uv pip install vllm --torch-backend=auto
vLLM installation ↗
Serve the model
OpenAI-compatible API at http://localhost:8000/v1.
$vllm serve yandex/AliceAI-Foundation-80B-A3B-Base --max-model-len 16384
Runs the original weights, not the Q4_K_M GGUF; needs ~179 GB of GPU memory.

macOS only.

MLX supports only macOS, so follow these steps for that setup.

Install mlx-lm
$pip install mlx-lm
Chat in the terminal
$mlx_lm.chat --model Yamada114514/AliceAI-Foundation-80B-A3B-Base-MLX-4bit
Downloads the model on first run. The MLX repo has its own quantisation; it may not match the quant selected above.
Or start a local server
Listens on http://localhost:8080.
$mlx_lm.server --model Yamada114514/AliceAI-Foundation-80B-A3B-Base-MLX-4bit

Works on macOS, Windows and Linux.

Install Jan
jan.ai ↗
Find the model
Open the Hub tab, paste this model ID in Search, and choose Q4_K_M if offered.
$Yamada114514/AliceAI-Foundation-80B-A3B-Base-GGUF
Use the local API (optional)
Settings > Local API Server > Start Server. It listens on http://127.0.0.1:1337.

Rent GPUs hourly to run Q4_K_M right there. Pricing comes from partner marketplaces and varies.

Available GPU rentals for Q4_K_M, prioritized by value
GPU	Speed	$/1M tokens	Rent
A100 80GBCheapestCheapest30 GB free	~194 tok/s	$1.62	Deploy on Vast.aifrom $1.14/hr (opens in new tab)	30 GB free · ~194 tok/s · $1.62 / 1M tokens
RTX PRO 6000 Max-Q (96 GB)46 GB free	~182 tok/s	$2.03	Deploy on Vast.aifrom $1.33/hr (opens in new tab)	96 GB · 46 GB free · ~182 tok/s · $2.03 / 1M tokens
A100 80GB30 GB free	~194 tok/s	$1.99	Deploy on RunPodfrom $1.39/hr (opens in new tab)	30 GB free · ~194 tok/s · $1.99 / 1M tokens
RTX PRO 6000 (96 GB)46 GB free	~182 tok/s	$2.34	Deploy on Vast.aifrom $1.54/hr (opens in new tab)	96 GB · 46 GB free · ~182 tok/s · $2.34 / 1M tokens
RTX PRO 6000 Server (96 GB)46 GB free	~171 tok/s	$2.74	Deploy on RunPodfrom $1.69/hr (opens in new tab)	96 GB · 46 GB free · ~171 tok/s · $2.74 / 1M tokens
RTX PRO 6000 Server (96 GB)46 GB free	~171 tok/s	$2.82	Deploy on Vast.aifrom $1.74/hr (opens in new tab)	96 GB · 46 GB free · ~171 tok/s · $2.82 / 1M tokens
H100 (80 GB)30 GB free	~240 tok/s	$3.11	Deploy on RunPodfrom $2.69/hr (opens in new tab)	80 GB · 30 GB free · ~240 tok/s · $3.11 / 1M tokens
H200 (141 GB)91 GB free	~270 tok/s	$3.69	Deploy on RunPodfrom $3.59/hr (opens in new tab)	141 GB · 91 GB free · ~270 tok/s · $3.69 / 1M tokens

Compare GPUs →

How Much VRAM Does AliceAI Foundation 80B A3B Base Need?

Select a quantization level; the listed hardware adheres to it.

VRAM	+32K	+128K	File	Quality	On RTX 4090 (reference)
35.0 GB	36.5 GB	41.3 GB	29.1 GB	2-bit · K-quant — 2-bit quantization with K-quant improvements	Does not fit: needs 35.7 GB
36.0 GB	37.5 GB	42.3 GB	34.5 GB	3-bit small — 3-bit small quantization	Does not fit: needs 36.7 GB
40.0 GB	41.5 GB	46.4 GB	~39.6 GB	3-bit medium — 3-bit medium quantization	Does not fit: needs 40.7 GB
42.1 GB	43.6 GB	48.4 GB	41.2 GB	3-bit large — 3-bit large quantization	Does not fit: needs 42.8 GB
46.1 GB	47.6 GB	52.5 GB	45.5 GB	4-bit small — 4-bit small quantization	Does not fit: needs 46.8 GB
49.2 GB	50.7 GB	55.5 GB	48.4 GB	4-bit medium — 4-bit medium quantization — most popular sweet spot	Does not fit: needs 49.9 GB
56.3 GB	57.8 GB	62.6 GB	55 GB	5-bit small — 5-bit small quantization	Does not fit: needs 57 GB
58.3 GB	59.8 GB	64.7 GB	56.7 GB	5-bit medium — 5-bit medium quantization — good quality/size tradeoff	Does not fit: needs 59 GB
67.5 GB	69 GB	73.8 GB	65.5 GB	6-bit, very good — 6-bit quantization, very good quality	Does not fit: needs 68.2 GB
81.7 GB	83.2 GB	88 GB	84.8 GB	8-bit, near-lossless — 8-bit quantization, near-lossless	Does not fit: needs 82.4 GB
163.0 GB	164.5 GB	169.3 GB	159.4 GB	16-bit · bfloat — Brain floating point 16 — preferred for training	Does not fit: needs 163.7 GB

est.calculated, no published file yet

Selected Q4_K_M

4-bit medium quantization

49.9GBat 16K context

Weights: 48.8 GB
KV cache: 0.8 GB
Overhead: 0.3 GB

Doesn't fit RTX 4090 (reference) · needs 25.9 GB more

≥50 tok/s needs ≥294 GB/s · M3 Ultra ~65, M2 Ultra ~64

Or .

Which Hardware Can Run AliceAI Foundation 80B A3B Base?

Q4_K_M · 49.2 GB

AliceAI Foundation 80B A3B Base (Q4_K_M) requires 49.2 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 64+ GB is recommended. Using the full 262K context window can add up to 12.8 GB, bringing total usage to 62.0 GB. No consumer GPU has enough memory. 52 devices with unified memory can run AliceAI Foundation 80B A3B Base, including Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB), Mac Pro M2 Ultra (64 GB).

By VRAM tier · typical GPUs and Macs at Q4_K_M

Typical hardwareQuantVRAMtok/sContext that fits

8–16 GB · RTX 4060, RTX 3060 12GB, RTX 5060 Ti 16GBDoesn't fit — smallest quant needs 35.7 GB8 GB tier →12 GB tier →16 GB tier →

24 GB · RTX 4090referenceDoesn't fit — smallest quant needs 35.7 GB24 GB tier →

Mac 32 GB · Mac Mini M4Doesn't fit — smallest quant needs 35.7 GBMacs →

Mac 64 GB · Mac Studio M4 MaxQ4_K_M · 49.9 GB · ~55 tok/s · 128K49.9~55128KMacs →

Alternatives to AliceAI Foundation 80B A3B Base

Similar models

3tok/s on RTX 4090

Model	Score	VRAM	tok/s	Why it is similar
NVIDIA Nemotron Labs 3 Puzzle 75B A9B BF16NVIDIA	—	54.5 GB	—needs > 24 GB	same size
Qwen3 Next 80B A3B InstructAlibaba	—	22.9 GB	~200	same size
Hunyuan A13B InstructTencent	—	50.7 GB	—needs > 24 GB	same size

Specifications

Publisher: yandex
Parameters: 81.3B
Active parameters: 5.5B per token
Architecture: AliceAIForCausalLM
Context Length: 262,144 tokens
Vocabulary Size: 129,024
Release Date: 2026-09-12
License: Apache 2.0

Resources

Hugging Face model card (opens in a new tab)

Frequently Asked Questions

How much VRAM does AliceAI Foundation 80B A3B Base need?

AliceAI Foundation 80B A3B Base requires 49.2 GB of VRAM at Q4_K_M, or 163.0 GB at BF16. Full 262K context adds up to 12.8 GB (62.0 GB total).

VRAM = Weights + KV Cache + Overhead

Weights = 81.3B × 4.8 bits ÷ 8 = 48.8 GB

KV Cache + Overhead ≈ 0.4 GB (at 2K context + ~0.3 GB framework)

Fit ratings and hardware model lists check this model with room for a 16K-token context, which needs a little more memory.

KV Cache + Overhead ≈ 13.2 GB (at full 262K context)

VRAM usage by quantization

Q4_K_M

49.2 GB

Q4_K_M + full context

62.0 GB

Learn more about VRAM estimation →

Can NVIDIA GeForce RTX 5090 run AliceAI Foundation 80B A3B Base?

No — AliceAI Foundation 80B A3B Base requires at least 35.0 GB at Q2_K, which exceeds the NVIDIA GeForce RTX 5090's 32 GB of VRAM.

What's the best quantization for AliceAI Foundation 80B A3B Base?

For AliceAI Foundation 80B A3B Base, Q4_K_M (49.2 GB) offers the best balance of quality and VRAM usage. Q5_K_S (56.3 GB) provides better quality if you have the VRAM. The smallest option is Q2_K at 35.0 GB.

VRAM requirement by quantization

Q2_K

35.0 GB~75%

Q3_K_L

42.1 GB~86%

Q4_K_M ★

49.2 GB~89%

Q5_K_S

56.3 GB~92%

Q6_K

67.5 GB~95%

BF16

163.0 GB~100%

★ Recommended — best balance of quality and VRAM usage.

Learn more about quantization →

Can I run AliceAI Foundation 80B A3B Base on a Mac?

Yes — Mac Studio M5 Max (48 GB) and 55 other Macs can run AliceAI Foundation 80B A3B Base, the smallest at Q3_K_L. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 49.2 GB of usable unified memory (RAM minus macOS overhead), from Mac Pro M2 Ultra (64 GB).

Can I run AliceAI Foundation 80B A3B Base locally?

Yes — AliceAI Foundation 80B A3B Base can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.

How fast is AliceAI Foundation 80B A3B Base?

At Q4_K_M, AliceAI Foundation 80B A3B Base can reach ~97 tok/s on AMD Instinct MI350X. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.

tok/s = 1000 ÷ (active GB ÷ (bandwidth GB/s × efficiency) × 1000 + layers × routing ms)

Mixture-of-Experts: just the active experts get read for each token, along with a fixed routing cost per layer.

Example: NVIDIA B200 → 3.3 GB active ÷ (8000 × 0.65) = 0.64 ms, plus 48 layers × 0.055 ms = 2.64 ms, so 1000 ÷ 3.28 ms = ~305 tok/s

Estimated speed at Q4_K_M (49.2 GB)

NVIDIA B200

~305 tok/s

NVIDIA B300

~305 tok/s

NVIDIA GH200 Grace Hopper Superchip

~272 tok/s

Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.

Learn more about tok/s estimation →

What's the download size of AliceAI Foundation 80B A3B Base?

At Q4_K_M, the download is about 48.77 GB. The full-precision BF16 version is 162.57 GB. The smallest option (Q2_K) is 34.55 GB.

Which GPUs can run AliceAI Foundation 80B A3B Base?

No single consumer GPU has enough VRAM to run AliceAI Foundation 80B A3B Base at Q4_K_M (49.2 GB). Multi-GPU or professional hardware is required.

Which devices can run AliceAI Foundation 80B A3B Base?

52 devices with unified memory can run AliceAI Foundation 80B A3B Base at Q4_K_M (49.2 GB), including ASUS Ascent GX10, Beelink GTR9 Pro (Ryzen AI Max+ 395, 128 GB), Framework Desktop (Ryzen AI Max+ 395, 128 GB), GMKtec EVO-X2 (Ryzen AI Max+ 395, 128 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.