NVIDIA Jetson AGX Thor Developer Kit

Hardware · NVIDIA

What AI models can the NVIDIA Jetson AGX Thor (128 GB) run?

Unified memory
128 GB
Usable for models
115 GB
Memory bandwidth
273 GB/s
Runs models on
GPU (CUDA)
Released
2025
548 / 622 LLMs run Best: Kolibri 1 (78B-A3.5B) 61 / 61 Image models run Best: MiniMax H3 Image 43 / 43 Video models run Best: LTX-2.5 (22B, video + audio) 70 / 70 Decision models run Best: Jina Reranker v3.5

Models run on the GPU (CUDA) here, sharing the board’s memory with the system. We plan around 115 GB of the 128 GB (what Linux leaves free for a model). Blackwell GPU with 2,560 CUDA cores and 96 fifth-generation Tensor Cores (up to 2,070 FP4 TFLOPS, sparse).

More about this hardware

MakerNVIDIA
TypeSingle-board computer or edge AI board
GPU share of memory90% (what the GPU can typically be given for a model)
ChipNVIDIA Jetson AGX Thor
AI acceleratorBlackwell GPU with 2,560 CUDA cores and 96 fifth-generation Tensor Cores (up to 2,070 FP4 TFLOPS, sparse).
Boards with this chipJetson AGX Thor Developer Kit · Jetson T5000 module
Memory type128 GB LPDDR5X, 256-bit
Bandwidth figure273.0 GB/s, as the maker states it
NoteUnified LPDDR5X shared with Jetson Linux; estimate ~10% OS reserve (no official cap).
Specificationswww.nvidia.com

Apps that run models well on the NVIDIA Jetson AGX Thor (128 GB)

Linux on ARM boards

  1. OllamaOne command with Ollama’s own build of this model.
  2. vLLMThe full model fits in GPU memory at 8-bit, so it can batch many requests: best for serving a team or an app.
  3. LM StudioPoint-and-click: search, download and chat in one window.
  4. llama.cppUsually the first to support new model architectures.

For a typical model that fits in memory. Each model page picks the best app for that model on this hardware. All apps

1 product with the NVIDIA Jetson AGX Thor (128 GB)

Best LLMs for the NVIDIA Jetson AGX Thor (128 GB)

Quality Model Quant Memory Speed Runs
91Excellent quality (A+) · 91 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 95/100 (quality, fit, speed and memory use) sets the ranking. Kolibri 1 (78B-A3.5B)Aleph Alpha Q8_0 92.8 GB 41.5 tok/s Runs great
86Excellent quality (A+) · 86 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 93/100 (quality, fit, speed and memory use) sets the ranking. Qwen3 Coder Next (80B-A3B)Alibaba Q8_0 94.5 GB 46.8 tok/s Runs great
86Excellent quality (A+) · 86 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 93/100 (quality, fit, speed and memory use) sets the ranking. Qwen3-Next 80B-A3B ThinkingAlibaba Q8_0 96.2 GB 46.8 tok/s Runs great
84Great quality (A) · 84 of 100: how capable the model is at MXFP4, whatever it runs on. Match score 92/100 (quality, fit, speed and memory use) sets the ranking. gpt-oss 120BOpenAI MXFP4 76.7 GB 56.9 tok/s Runs great
84Great quality (A) · 84 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 92/100 (quality, fit, speed and memory use) sets the ranking. Qwen3 Next 80B-A3B InstructAlibaba Q8_0 96.2 GB 46.8 tok/s Runs great
88Excellent quality (A+) · 88 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 92/100 (quality, fit, speed and memory use) sets the ranking. Ornith 1.5 122B-A10BDeepReinforce Q4_K_M 86.3 GB 30.1 tok/s Runs great
87Excellent quality (A+) · 87 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 92/100 (quality, fit, speed and memory use) sets the ranking. Qwen3.5 122B-A10BAlibaba Q4_K_M 86.3 GB 30.1 tok/s Runs great
89Excellent quality (A+) · 89 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 92/100 (quality, fit, speed and memory use) sets the ranking. Nemotron 3.5 Super VL 120B-A12BNVIDIA Q4_K_M 85.8 GB 25.5 tok/s Runs well
93Excellent quality (A+) · 93 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. Qwen3.6 35B-A3BAlibaba Q8_0 48.1 GB 46.8 tok/s Runs great
93Excellent quality (A+) · 93 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. Qwen3.5 35B-A3BAlibaba Q8_0 48.1 GB 46.8 tok/s Runs great
93Excellent quality (A+) · 93 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. MiniCPM-V 4.7 35B-A3BOpenBMB Q8_0 47.3 GB 46.8 tok/s Runs great
81Great quality (A) · 81 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. Laguna S 2.1 (118B-A8B)Poolside Q4_K_M 83.1 GB 36.7 tok/s Runs great

Best Image models for the NVIDIA Jetson AGX Thor (128 GB)

Quality Model Quant Memory Speed Runs
91Excellent quality (A+) · 91 of 100: how capable the model is at FP8, whatever it runs on. Match score 87/100 (quality, fit, speed and memory use) sets the ranking. MiniMax H3 ImageMiniMax FP8 72.0 GB 52 s Slow
84Great quality (A) · 84 of 100: how capable the model is at FP16, whatever it runs on. Match score 84/100 (quality, fit, speed and memory use) sets the ranking. FLUX.2 [klein] 9BBlack Forest Labs FP16 37.1 GB 10 s Runs great
92Excellent quality (A+) · 92 of 100: how capable the model is at FP16, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking. Krea 2 TurboKrea FP16 37.4 GB 25 s Runs well
90Excellent quality (A+) · 90 of 100: how capable the model is at FP8, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. Cosmos3-Super Text2Image (64B)NVIDIA FP8 69.1 GB 3.5 min Very slow
88Excellent quality (A+) · 88 of 100: how capable the model is at FP8, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. HunyuanImage 3.0 (80B-A13B)Tencent FP8 86.9 GB 5.2 min Very slow
91Excellent quality (A+) · 91 of 100: how capable the model is at Q8 GGUF, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. FLUX.2 [dev]Black Forest Labs Q8 GGUF 61.9 GB 4.0 min Very slow

Best Video models for the NVIDIA Jetson AGX Thor (128 GB)

Quality Model Quant Memory Speed Runs
88Excellent quality (A+) · 88 of 100: how capable the model is at FP16, whatever it runs on. Match score 88/100 (quality, fit, speed and memory use) sets the ranking. LTX-2.5 (22B, video + audio)Lightricks FP16 82.5 GB 8.7 min Runs well
86Excellent quality (A+) · 86 of 100: how capable the model is at FP16, whatever it runs on. Match score 82/100 (quality, fit, speed and memory use) sets the ranking. LTX-2.3Lightricks FP16 75.4 GB 38.4 min Slow
89Excellent quality (A+) · 89 of 100: how capable the model is at Q8 GGUF, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. MiniMax H3 (33B, video + audio)MiniMax Q8 GGUF 68.2 GB 1.9 h Very slow
88Excellent quality (A+) · 88 of 100: how capable the model is at FP16, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. Kandinsky 6.0 Pro (29B, video + audio)Sber AI FP16 85.4 GB 4.7 h Very slow
82Great quality (A) · 82 of 100: how capable the model is at FP16, whatever it runs on. Match score 81/100 (quality, fit, speed and memory use) sets the ranking. LTX-2 (19B, video + audio)Lightricks FP16 75.4 GB 38.4 min Slow

Best Decision models for the NVIDIA Jetson AGX Thor (128 GB)

Quality Model Quant Memory Speed Runs
86Excellent quality (A+) · 86 of 100: how capable the model is at F16, whatever it runs on. Match score 93/100 (quality, fit, speed and memory use) sets the ranking. Jina Reranker v3.5Jina AI F16 11.3 GB 65 ms Runs great
84Great quality (A) · 84 of 100: how capable the model is at F16, whatever it runs on. Match score 92/100 (quality, fit, speed and memory use) sets the ranking. Jina Reranker v3Jina AI F16 11.3 GB 65 ms Runs great
80Great quality (A) · 80 of 100: how capable the model is at F16, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. FRIDA-DecisionsSber AI F16 2.4 GB 50 ms Runs great
80Great quality (A) · 80 of 100: how capable the model is at F16, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. GLiNER2.5 DecideFastino F16 1.7 GB 31 ms Runs great
80Great quality (A) · 80 of 100: how capable the model is at F16, whatever it runs on. Match score 91/100 (quality, fit, speed and memory use) sets the ranking. Qwen3 Reranker 0.6BAlibaba F16 11.3 GB 65 ms Runs great
76Great quality (A) · 76 of 100: how capable the model is at F16, whatever it runs on. Match score 89/100 (quality, fit, speed and memory use) sets the ranking. Llama Prompt Guard 2 86MMeta F16 1.2 GB 35 ms Runs great
76Great quality (A) · 76 of 100: how capable the model is at F16, whatever it runs on. Match score 89/100 (quality, fit, speed and memory use) sets the ranking. mxbai-rerank-base v2Mixedbread F16 10.8 GB 54 ms Runs great
82Great quality (A) · 82 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 89/100 (quality, fit, speed and memory use) sets the ranking. mxbai-rerank-large v2Mixedbread Q8_0 11.5 GB 150 ms Runs well

Best LLMs by task on this hardware Open in Model Finder

Estimates come from published hardware specs and our open formulas, not from running the model on your machine. Real speed varies with runtime, drivers and settings. Numbers marked “Measured” come from a linked benchmark.

Was this page helpful?