Recurrentgemma 2B Flax — Hardware Requirements & GPU Compatibility
Chat
Recurrentgemma 2B Flax is a 2B-parameter open language model from Google in the Gemma 2 family. At BF16 it needs about 4.40 GB of VRAM — see which GPUs and Macs can run it below.
1 downloads
5 likesSpecifications
- Publisher
- Family
- Gemma 2
- Parameters
- 2B
- Release Date
- 2024-06-27
- License
- Gemma Terms
Get Started
HuggingFace
How Much VRAM Does Recurrentgemma 2B Flax Need?
Select a quantization to see compatible GPUs below.
| Quantization | Bits | VRAM | + Context | File Size | Quality |
|---|---|---|---|---|---|
| BF16 | 16.00 | 4.4 GB | — | 4.00 GB | Brain floating point 16 — preferred for training |
Which GPUs Can Run Recurrentgemma 2B Flax?
BF16 · 4.4 GBRecurrentgemma 2B Flax (BF16) requires 4.4 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 6+ GB is recommended. 35 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti.
Runs great
— Plenty of headroomWhich Devices Can Run Recurrentgemma 2B Flax?
BF16 · 4.4 GB33 devices with unified memory can run Recurrentgemma 2B Flax, including NVIDIA DGX H100, NVIDIA DGX A100 640GB.
Runs great
— Plenty of headroomRelated Models
Frequently Asked Questions
- How much VRAM does Recurrentgemma 2B Flax need?
Recurrentgemma 2B Flax requires 4.4 GB of VRAM at BF16.
VRAM = Weights + KV Cache + Overhead
Weights = 2B × 16 bits ÷ 8 = 4 GB
KV Cache + Overhead ≈ 0.4 GB (at 2K context + ~0.3 GB framework)
VRAM usage by quantization
BF164.4 GB- Can I run Recurrentgemma 2B Flax on a Mac?
Recurrentgemma 2B Flax requires at least 4.4 GB at BF16, which exceeds the unified memory of most consumer Macs. You would need a Mac Studio or Mac Pro with a high-memory configuration.
- Can I run Recurrentgemma 2B Flax locally?
Yes — Recurrentgemma 2B Flax can run locally on consumer hardware. At BF16 quantization it needs 4.4 GB of VRAM. Popular tools include Ollama, LM Studio, and llama.cpp.
- How fast is Recurrentgemma 2B Flax?
At BF16, Recurrentgemma 2B Flax can reach ~663 tok/s on AMD Instinct MI300X. On NVIDIA GeForce RTX 4090: ~149 tok/s. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.
tok/s = (bandwidth GB/s ÷ model GB) × efficiency
Example: AMD Instinct MI300X → 5300 ÷ 4.4 × 0.55 = ~663 tok/s
Estimated speed at BF16 (4.4 GB)
~663 tok/s~149 tok/s~495 tok/s~410 tok/sReal-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
- What's the download size of Recurrentgemma 2B Flax?
At BF16, the download is about 4.00 GB.
- Which GPUs can run Recurrentgemma 2B Flax?
35 consumer GPUs can run Recurrentgemma 2B Flax at BF16 (4.4 GB). Top options include AMD Radeon RX 6700 XT, AMD Radeon RX 6800, AMD Radeon RX 6800 XT. 35 GPUs have plenty of headroom for comfortable inference.
- Which devices can run Recurrentgemma 2B Flax?
33 devices with unified memory can run Recurrentgemma 2B Flax at BF16 (4.4 GB), including Mac Mini M4 (16 GB), Mac Mini M4 (32 GB), Mac Mini M4 Pro (24 GB), Mac Mini M4 Pro (48 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.