LLaDA2.2 Flash — Hardware Requirements & GPU Compatibility
Chat
LLaDA2.2 Flash is a 102.9B-parameter open language model from Inclusion AI. It supports a context window of up to 131,072 tokens. At BF16 it needs about 206.21 GB of VRAM — see which GPUs and Macs can run it below.
LLaDA2.2 Flash runs on a 256 GB+ Mac or a datacenter GPU
Best local pick → Mac Studio (M3 Ultra, 256GB) · BF16 · 207.2 GB · ~3 tok/s
Macs from 256 GB (BF16 · ~3 tok/s)
Apache 2.0
Not set up for it?
Reference GPU · RTX 4090 24 GB
Estimates, ±20% · how we estimate
Run LLaDA2.2 Flash
One recommended way. Every other method, OS and quant is in the step-by-step guide.
Recommended
vllm serve inclusionAI/LLaDA2.2-flash --max-model-len 16384 --tensor-parallel-size 8vLLM repo with its own quantisation
Download: Weights on Hugging Face
Other ways
each opens step-by-stepRun in cloud
Fits on B300 (288 GB) (80 GB headroom) · BF16
- Generation speed
- ~25 tok/s
- generation speed
- Cost per 1M output tokens
- $76.50
- per 1M output tokens
How Much VRAM Does LLaDA2.2 Flash Need?
Select a quantization level; the listed hardware adheres to it.
| Quantization | VRAM | +32K | Runs on | |
|---|---|---|---|---|
| 206.2 GB | 208.2 GB | 16-bit · bfloat — Brain floating point 16 — preferred for training | Runs on Instinct MI325X 256 GB |
Selected BF16
Brain floating point 16207.2GBat 16K context
- Weights
- 205.8 GB
- KV cache
- 1.1 GB
- Overhead
- 0.3 GB
Server-class: ~207.2 GB at 16K context · one Instinct MI325X 256 GB
Which Hardware Can Run LLaDA2.2 Flash?
BF16 · 206.2 GBLLaDA2.2 Flash (BF16) requires 206.2 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 269+ GB is recommended. Using the full 131K context window can add up to 8.5 GB, bringing total usage to 214.7 GB. No consumer GPU has enough memory. 2 devices with unified memory can run LLaDA2.2 Flash, including Mac Studio (M3 Ultra, 512GB), Mac Studio (M3 Ultra, 256GB).
Needs server-class hardware
LLaDA2.2 Flash runs on a 256 GB+ Mac or a datacenter GPU · BF16 needs ~207.2 GB
| Hardware | Quant | Memory | Speed | Context | Hardware page | |
|---|---|---|---|---|---|---|
| Mac 256 GB · Mac Studio (M3 Ultra, 256GB) | BF16 | 207.2 GB | ~3 tok/s | 128K ctx | Mac Studio → | BF16 · 207.2 GB · ~3 tok/s · 128K ctx |
| Instinct MI325X 256 GB · one GPU | BF16 | 207.2 GB | ~18 tok/s | 128K ctx | Instinct MI325X → | BF16 · 207.2 GB · ~18 tok/s · 128K ctx |
| 3× H100 SXM 80 GB · 240 GB | BF16 | 207.2 GB | — | 128K ctx | H100 SXM → | BF16 · 207.2 GB · 128K ctx |
| 2× H200 SXM 141 GB · 282 GB | BF16 | 207.2 GB | — | 128K ctx | H200 SXM → | BF16 · 207.2 GB · 128K ctx |
| 2× B200 192 GB · 384 GB | BF16 | 207.2 GB | — | 128K ctx | B200 → | BF16 · 207.2 GB · 128K ctx |
Also: Instinct MI325X · Instinct MI350X · Instinct MI355X · B300 · Mac Studio (M3 Ultra, 256GB) · Mac Studio (M3 Ultra, 512GB) · NVIDIA DGX A100 640GB · NVIDIA DGX H100
By VRAM tier · typical GPUs and Macs at BF16
Consumer GPUs (up to 32 GB) and Macs up to 192 GB: too small for this model (needs ≥ 207.2 GB)8 GB tier →12 GB tier →16 GB tier →24 GB tier →Macs →
Similar models
4| Model | Score | VRAM | tok/s | Why it is similar |
|---|---|---|---|---|
| Llama 4 Scout 17B 16E InstructMeta | 81.8 | 77 GB | —needs > 24 GB | same size |
| C4ai Command A 03 2025Cohere | 79.8 | 78.7 GB | —needs > 24 GB | same size |
| GLM 4.6VZ.ai | — | 66 GB | —needs > 24 GB | same size |
| C4ai Command R Plus 08 2024Cohere | — | 73.8 GB | —needs > 24 GB | same size |
Specifications
- Publisher
- Inclusion AI
- Parameters
- 102.9B
- Architecture
- LLaDA2MoeModelLM
- Context Length
- 131,072 tokens
- Vocabulary Size
- 157,184
- Release Date
- 2026-07-16
- License
- Apache 2.0
Frequently Asked Questions
How much VRAM does LLaDA2.2 Flash need?
LLaDA2.2 Flash requires 206.2 GB of VRAM at BF16. Full 131K context adds up to 8.5 GB (214.7 GB total).
VRAM = Weights + KV Cache + Overhead
Weights = 102.9B × 16 bits ÷ 8 = 205.8 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 ≈ 8.9 GB (at full 131K context)
VRAM usage by quantization
Can NVIDIA GeForce RTX 5090 run LLaDA2.2 Flash?
No — LLaDA2.2 Flash requires at least 206.2 GB at BF16, which exceeds the NVIDIA GeForce RTX 5090's 32 GB of VRAM.
Can I run LLaDA2.2 Flash on a Mac?
Yes — Mac Studio (M3 Ultra, 256GB) and 1 other Mac can run LLaDA2.2 Flash. Apple Silicon uses unified memory, so the model shares RAM with the system. At BF16 you need at least 206.2 GB of usable unified memory (RAM minus macOS overhead).
Can I run LLaDA2.2 Flash locally?
Yes — LLaDA2.2 Flash can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.
How fast is LLaDA2.2 Flash?
At BF16, LLaDA2.2 Flash can reach ~23 tok/s on AMD Instinct MI350X. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.
tok/s = (bandwidth GB/s ÷ model GB) × efficiency
Example: NVIDIA B300 → 8000 ÷ 206.2 × 0.65 = ~25 tok/s
Estimated speed at BF16 (206.2 GB)
Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
What's the download size of LLaDA2.2 Flash?
At BF16, the download is about 205.78 GB.
Which GPUs can run LLaDA2.2 Flash?
No single consumer GPU has enough VRAM to run LLaDA2.2 Flash at BF16 (206.2 GB). Multi-GPU or professional hardware is required.
Which devices can run LLaDA2.2 Flash?
2 devices with unified memory can run LLaDA2.2 Flash at BF16 (206.2 GB), including Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.