NVIDIAOrinAI Box

Best AI Models for NVIDIA Jetson AGX Orin 64GB

Memory:64 GB Unified·Bandwidth:204.8 GB/s·GPU Cores:2048 GPU cores·CPU Cores:12 CPU cores·Neural Engine:275.0 TOPS

64 GB unified − 3.5 GB OS overhead = 60.5 GB available for AI models

With 64 GB of memory, this is a high-end configuration for local AI. You can comfortably run most open-source LLMs including large 70B parameter models at good quantization levels, making it one of the best setups for serious local AI work.

At this memory tier, nearly every popular open-source model is within reach. You can run Llama 3 70B at Q4_K_M or even Q5_K_M quantization with room to spare, handle coding assistants like DeepSeek Coder 33B at high quality, and easily run any 7B–30B model at full or near-full precision. Context windows remain generous even with larger models, so multi-turn conversations and long-document processing work smoothly.

Runs Well

  • 70B models (Llama 3 70B, Qwen 72B) at Q4–Q5
  • 30B models at Q6–Q8 quality
  • 7B–14B models at full FP16 precision
  • Vision models (LLaVA, CogVLM) without compromise

Challenging

  • Mixture-of-experts models like Mixtral 8x22B at higher quants
  • 120B+ models still require lower quantizations

What LLMs Can NVIDIA Jetson AGX Orin 64GB Run?

275 models · 46 excellent · 91 good

Showing compatibility for NVIDIA Jetson AGX Orin 64GB

LLM models compatible with NVIDIA Jetson AGX Orin 64GB — ranked by performance
ModelVRAMGrade
Q4_K_M·48.9 t/s tok/s·262K ctx·RUNS WELL
16.1 GBA79
Q4_K_M·60.1 t/s tok/s·262K ctx·RUNS WELL
19.3 GBA84
Q4_K_M·49.7 t/s tok/s·262K ctx·RUNS WELL
21.4 GBA79
Qwen3.8 27B27.8B
Q4_K_M·7.6 t/s tok/s·262K ctx·DECENT
17.4 GBB45
Muse Glimmer 30B29.8B
Q4_K_M·7.3 t/s tok/s·131K ctx·DECENT
18.3 GBB45
Qwen AgentWorld 35B A3B34.7B
Q4_K_M·55.4 t/s tok/s·262K ctx·RUNS WELL
21.2 GBA82
North Mini Code 1.030.5B
Q4_K_M·55.8 t/s tok/s·500K ctx·RUNS WELL
18.7 GBA82
Q4_K_M·48.9 t/s tok/s·262K ctx·RUNS WELL
16.1 GBA79
Q4_K_M·153.4 t/s tok/s·128K ctx·RUNS GREAT
5.5 GBS97
GLM 4.7 Flash31.2B
Q4_K_M·40.7 t/s tok/s·203K ctx·RUNS WELL
19.8 GBA75
Q4_K_M·41.7 t/s tok/s·262K ctx·RUNS WELL
21.9 GBA76
Q4_K_M·36.0 t/s tok/s·262K ctx·RUNS WELL
49.2 GBA73
Q4_K_M·119.9 t/s tok/s·131K ctx·RUNS GREAT
5.3 GBS95
Q4_K_M·76.4 t/s tok/s·131K ctx·RUNS GREAT
7.7 GBS87
Agents A135.1B
Q4_K_M·49.7 t/s tok/s·262K ctx·RUNS WELL
21.4 GBA79
Unlimited OCR3.3B
Q4_K_M·144.9 t/s tok/s·33K ctx·RUNS GREAT
2.4 GBS96

NVIDIA Jetson AGX Orin 64GB Specifications

Brand
NVIDIA
Chip
Orin
Type
AI Box
Unified Memory
64 GB
Memory Bandwidth
204.8 GB/s
GPU Cores
2048
CPU Cores
12
Neural Engine
275.0 TOPS
Architecture
Ampere
Memory Type
LPDDR5
TDP
60 W
Release Date
2022-03-22

Get Started

Ollama (Recommended)

$curl -fsSL https://ollama.com/install.sh | sh
$ollama run llama3:8b

LM Studio

LM Studio

Download LM Studio, search for a model, and run it with one click.

Prompt Processing

Estimated for Qwen3 Coder Next, the compute-bound phase that reads your prompt before the first reply token appears.

2,467.8tok/s prefill

Short chat

207 ms

512 tok prompt

Long chat

1.7 s

4,096 tok prompt

Document / codebase

13.3 s

32,768 tok prompt

Prefill is compute-bound and a different number from the decode tok/s shown elsewhere on this page — how prompt processing works →

Performance figures are estimates calibrated as of 2026-09-21 — see calibration basis →

Devices to Consider

Similar devices and upgrades with more memory or higher bandwidth

Frequently Asked Questions

Can NVIDIA Jetson AGX Orin 64GB run Gemma 4 26B A4B IT?

Yes, the NVIDIA Jetson AGX Orin 64GB with 64 GB unified memory can run Gemma 4 26B A4B IT, NVIDIA Nemotron 3.5 Lightning 30B A3B BF16, Ornith 1.0 35B, and 2535 other models. 722 models achieve excellent performance, and 782 run at good quality. Apple Silicon's unified memory architecture lets the GPU access the full memory pool without copying data, making it efficient for AI workloads.

How much memory is available for AI on NVIDIA Jetson AGX Orin 64GB?

The NVIDIA Jetson AGX Orin 64GB has 64 GB unified memory. After macOS reserves ~3.5 GB for the operating system, approximately 60.5 GB is available for AI models. Unlike discrete GPUs where VRAM is separate from system RAM, Apple Silicon shares one memory pool between the CPU and GPU — this means no data copying overhead, but you share memory with macOS and open apps.

Is NVIDIA Jetson AGX Orin 64GB good for AI?

With 64 GB unified memory and 204.8 GB/s bandwidth, the NVIDIA Jetson AGX Orin 64GB is excellent for running local AI models. It supports 1504 models at good quality or better. This is a premium configuration — you can run large 30B+ parameter models at good quality, and most 7B models at maximum quality. Ideal for professional AI workloads.

What's the best model for NVIDIA Jetson AGX Orin 64GB?

The top-rated models for the NVIDIA Jetson AGX Orin 64GB are Gemma 4 26B A4B IT, NVIDIA Nemotron 3.5 Lightning 30B A3B BF16, Ornith 1.0 35B. With this much memory, you can prioritize quality — use higher quantizations (Q5/Q6) for better output, or run larger 30B+ models for more capable reasoning.

How fast is NVIDIA Jetson AGX Orin 64GB for AI inference?

With 204.8 GB/s memory bandwidth, the NVIDIA Jetson AGX Orin 64GB achieves approximately 32 tok/s on a 7B model at Q4_K_M — that's comfortable for real-time chat. A 14B model runs at ~16 tok/s. Apple Silicon achieves high efficiency (~70%) thanks to unified memory — there's no PCIe bottleneck between CPU and GPU.

tok/s = (204.8 GB/s ÷ model GB) × efficiency

Apple Silicon achieves ~70% bandwidth efficiency thanks to unified memory and Metal acceleration.

Estimated speed on NVIDIA Jetson AGX Orin 64GB

Real-world results typically within ±20%.

Learn more about tok/s estimation →

Can I run AI offline on NVIDIA Jetson AGX Orin 64GB?

Yes — once you download a model, it runs entirely on the NVIDIA Jetson AGX Orin 64GB without internet. Applications like Ollama and LM Studio make it straightforward to download, manage, and run models locally. All your conversations stay private on your device with zero data sent to external servers. This is one of the key advantages of local AI: complete privacy, no API costs, and no rate limits.