Datagemma Rag 27B IT — Hardware Requirements & GPU Compatibility
Chat
Datagemma Rag 27B IT is a 27.2B-parameter open language model from Google in the Gemma family. At BF16 it needs about 59.90 GB of VRAM — see which GPUs and Macs can run it below.
Datagemma Rag 27B IT runs on a 64 GB+ Mac or a workstation GPU
Best local pick → Mac Pro M2 Ultra (64 GB) · BF16 · 63.5 GB · ~9 tok/s
Macs from 64 GB (BF16 · ~9 tok/s)
Gemma Terms
Not set up for it?
Reference GPU · RTX 4090 24 GB
Estimates, ±20% · how we estimate
Run Datagemma Rag 27B IT
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.
Recommended
vllm serve google/datagemma-rag-27b-it --max-model-len 16384vLLM repo with its own quantisation
Download: Weights on Hugging Face
Ways to runOther ways
each opens step-by-stepRun in cloud
Runs on A100 80GB (16 GB headroom) · BF16
- Generation speed
- ~22 tok/s
- generation speed
- Cost per 1M output tokens
- $14.11
- per 1M output tokens
How Much VRAM Does Datagemma Rag 27B IT Need?
Select a quantization level; the listed hardware adheres to it.
| Quantization | VRAM | +32K | On RTX 4090 (reference) | |
|---|---|---|---|---|
| 59.9 GB | 67.7 GB | 16-bit · bfloat — Brain floating point 16 — preferred for training | Does not fit: needs 63.5 GB |
Selected BF16
Brain floating point 1663.5GBat 16K context
- Weights
- 59.1 GB
- KV cache
- 4.1 GB
- Overhead
- 0.3 GB
Doesn't fit RTX 4090 (reference) · needs 39.5 GB more
Or .
Which Hardware Can Run Datagemma Rag 27B IT?
BF16 · 59.9 GBDatagemma Rag 27B IT (BF16) requires 59.9 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 78+ GB is recommended. No consumer GPU has enough memory. 52 devices with unified memory can run Datagemma Rag 27B IT, including Mac Studio (M3 Ultra, 256GB), Mac Studio (M3 Ultra, 512GB), Mac Pro M2 Ultra (64 GB).
64–80 GB
fits · 0.5 GB free · up to 16K ctx
3 Mac classes · 1 device family
M2 Ultra*64 GB~9Simulate output on Mac Pro M2 Ultra (64 GB)M5 Max64 GB~7Simulate output on Mac Studio M5 Max (64 GB)Orin~2Simulate output on NVIDIA Jetson AGX Orin 64GB
By VRAM tier · typical GPUs and Macs at BF16
Alternatives to Datagemma Rag 27B IT
Gemma family · 17 models →Similar models
6| Model | Score | VRAM | tok/s | Why it is similar |
|---|---|---|---|---|
| Qwen3.8 27BAlibaba | 122.0 | 17.9 GB | ~38 | same size |
| Qwen3.6 27BAlibaba | 111.2 | 17.9 GB | ~38 | same size |
| Qwen3.5 27BAlibaba | 103.8 | 17.9 GB | ~38 | same size |
| ERNIE 4.5 VL 28B A3B PTBaidu | — | 60 GB | —needs > 24 GB | same size |
| Ariarhymes-ai | — | 20.2 GB | ~168 | same size |
| JiRackDeltaNet 27BCMSManhattan | — | 17.6 GB | ~38 | same size |
Specifications
- Publisher
- Family
- Gemma
- Parameters
- 27.2B
- Release Date
- 2024-08-26
- License
- Gemma Terms
Frequently Asked Questions
How much VRAM does Datagemma Rag 27B IT need?
Datagemma Rag 27B IT requires 59.9 GB of VRAM at BF16.
VRAM = Weights + KV Cache + Overhead
Weights = 27.2B × 16 bits ÷ 8 = 54.5 GB
KV Cache + Overhead ≈ 5.4 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.
VRAM usage by quantization
Can NVIDIA GeForce RTX 5090 run Datagemma Rag 27B IT?
No — Datagemma Rag 27B IT requires at least 59.9 GB at BF16, which exceeds the NVIDIA GeForce RTX 5090's 32 GB of VRAM.
Can I run Datagemma Rag 27B IT on a Mac?
Yes — Mac Pro M2 Ultra (64 GB) and 41 other Macs can run Datagemma Rag 27B IT. Apple Silicon uses unified memory, so the model shares RAM with the system. At BF16 you need at least 59.9 GB of usable unified memory (RAM minus macOS overhead).
Can I run Datagemma Rag 27B IT locally?
Yes — Datagemma Rag 27B IT can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.
How fast is Datagemma Rag 27B IT?
At BF16, Datagemma Rag 27B IT can reach ~80 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 B200 → 8000 ÷ 59.9 × 0.65 = ~87 tok/s
Estimated speed at BF16 (59.9 GB)
Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
What's the download size of Datagemma Rag 27B IT?
At BF16, the download is about 54.45 GB.
Which GPUs can run Datagemma Rag 27B IT?
No single consumer GPU has enough VRAM to run Datagemma Rag 27B IT at BF16 (59.9 GB). Multi-GPU or professional hardware is required.
Which devices can run Datagemma Rag 27B IT?
52 devices with unified memory can run Datagemma Rag 27B IT at BF16 (59.9 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.