Cosmos Reason2 32B — Hardware Requirements & GPU Compatibility
VisionChatReasoningFunctionsCosmos Reason 2 32B is NVIDIA's 32-billion-parameter reasoning vision-language model for physical AI and robotics, built on Qwen3-VL-32B-Instruct. Given a text prompt and an image or video, it thinks through the answer using physics understanding and common sense, and the card describes uses such as robot planning, video analytics and dataset annotation, with improved spatio-temporal understanding and timestamp precision over the first generation. At this size it needs a high-end GPU with 24 GB or more of VRAM, or a unified-memory machine, once quantized. The context window is 262,144 tokens, and the card cites input of up to 256K tokens. It is released under the NVIDIA Open Model License, which the card describes as allowing commercial use and derivative models, with additional conditions; the repository uses gated access. Published in April 2026, it sits alongside 2B and 8B siblings in the Cosmos-Reason2 family.
Decent on a 24 GB GPU
Runs from 16 GB (IQ3_XXS · 16K)Best on 24 GB (Q4_K_M · ~32 tok/s · 16K)Macs from 16 GB (IQ2_XXS · ~14 tok/s)
Other
Reference GPU · RTX 4090 24 GB
Estimates, ±20% · how we estimate
Run Cosmos Reason2 32B
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
ollama run hf.co/mradermacher/Cosmos-Reason2-32B-i1-GGUF:Q4_K_MQ4_K_M fits fully on the RTX 4090 · 1.0 GB free · 16K context
Download: Cosmos-Reason2-32B.i1-Q4_K_M.gguf19.76 GBmradermacher
Ways to runOther ways
each opens step-by-stepRun in cloud
Your own private copy · Q4_K_M · 32 GB GPU
- The easiest way to try this model
- Running in a few minutes, easy setup
- Chat right here or connect any app
$5+ bonus for new users
Runs on RTX 5090 (32 GB) (9 GB headroom) · Q4_K_M
- Generation speed
- ~56 tok/s
- generation speed
- Cost per 1M output tokens
- $2.65
- per 1M output tokens
How Much VRAM Does Cosmos Reason2 32B Need?
Select a quantization level; the listed hardware adheres to it.
| Quantization | VRAM | +32K | On RTX 4090 (reference) | |
|---|---|---|---|---|
| 14.8 GB | 19.8 GB | 2-bit · K-quant — 2-bit quantization with K-quant improvements | Fits: 32K context · ~44 tok/s | |
| 15.2 GB | 20.3 GB | 3-bit small — 3-bit small quantization | Fits: 32K context · ~43 tok/s | |
| 16.9 GB | 21.9 GB | 3-bit medium — 3-bit medium quantization | Fits: 32K context · ~39 tok/s | |
| 17.3 GB | 22.3 GB | 4-bit · legacy — 4-bit legacy quantization | Fits: 32K context · ~38 tok/s | |
| 20.6 GB | 25.7 GB | 4-bit medium — 4-bit medium quantization — most popular sweet spot | Fits: 16K context · ~32 tok/s | |
| 24.4 GB | 29.4 GB | 5-bit medium — 5-bit medium quantization — good quality/size tradeoff | Does not fit: needs 26.8 GB | |
| 28.2 GB | 33.2 GB | 6-bit, very good — 6-bit quantization, very good quality | Does not fit: needs 30.5 GB | |
| 34.0 GB | 39 GB | 8-bit, near-lossless — 8-bit quantization, near-lossless | Does not fit: needs 36.3 GB |
Selected Q4_K_M
4-bit medium quantization23.0GBat 16K context
- Weights
- 20.0 GB
- KV cache
- 2.7 GB
- Overhead
- 0.3 GB
Fits RTX 4090 (reference) · 1.0 GB free · ~32 tok/s
≥50 tok/s needs ≥1.59 TB/s · RTX 5090 ~56
Which Hardware Can Run Cosmos Reason2 32B?
Q4_K_M · 20.6 GBCosmos Reason2 32B (Q4_K_M) requires 20.6 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 27+ GB is recommended. Using the full 262K context window can add up to 42.6 GB, bringing total usage to 63.3 GB. 7 GPUs can run it, including NVIDIA GeForce RTX 5090, NVIDIA GeForce RTX 3090 Ti. 112 devices with unified memory can run Cosmos Reason2 32B, including Mac Studio M5 Ultra (256 GB), Mac Studio M5 Ultra (512 GB), Mac Studio M1 Max (32 GB).
20 GB
Q4_K_M · best fit Q4_0 (19.7 GB)
1 GPU
RX 7900 XT~28Simulate output on RX 7900 XT32–40 GB
fits · 9.0 GB free · up to 64K ctx
1 GPU · 3 Mac classes · 9 device families
RTX 5090~56Simulate output on RTX 5090M5 Max*36 GB~16Simulate output on Mac Studio M5 Max (36 GB)M632 GB~6Simulate output on iMac 24" M6 (32 GB)RTX Spark N1X~10Simulate output on NVIDIA RTX Spark N1X Laptop (32 GB)Orin~6Simulate output on NVIDIA Jetson AGX Orin 32GB
By VRAM tier · typical GPUs and Macs at Q4_K_M
Similar models
6| Model | Score | VRAM | tok/s | Why it is similar |
|---|---|---|---|---|
| Qwen3.6 35B A3BAlibaba | 106.8 | 22 GB | ~151 | same size |
| MiniCPM V 4.7 35B A3BOpenBMB | — | 22.1 GB | ~169 | same size |
| Ornith 1.5 35B A3Bornith-ai | — | 22 GB | ~151 | same size |
| Agents A1InternScience | — | 21.5 GB | ~171 | same size |
| EXAONE 4.5 33BLGAI-EXAONE | — | 23.9 GB | ~33 | same size |
| Gemma 4 31BGoogle | — | 21.7 GB | ~32 | same size |
Specifications
- Publisher
- NVIDIA
- Parameters
- 33.4B
- Architecture
- Qwen3VLForConditionalGeneration
- Context Length
- 262,144 tokens
- Vocabulary Size
- 151,936
- Release Date
- 2026-04-29
- License
- Other
Frequently Asked Questions
How much VRAM does Cosmos Reason2 32B need?
Cosmos Reason2 32B requires 20.6 GB of VRAM at Q4_K_M, or 67.3 GB at BF16. Full 262K context adds up to 42.6 GB (63.3 GB total).
VRAM = Weights + KV Cache + Overhead
Weights = 33.4B × 4.8 bits ÷ 8 = 20 GB
KV Cache + Overhead ≈ 0.6 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 ≈ 43.3 GB (at full 262K context)
VRAM usage by quantization
Can NVIDIA GeForce RTX 4090 run Cosmos Reason2 32B?
Yes, at Q5_K_S (23.6 GB) or lower. Higher quantizations like Q5_K_M (24.4 GB) exceed the NVIDIA GeForce RTX 4090's 24 GB.
What's the best quantization for Cosmos Reason2 32B?
For Cosmos Reason2 32B, Q4_K_M (20.6 GB) offers the best balance of quality and VRAM usage. Q5_K_S (23.6 GB) provides better quality if you have the VRAM. The smallest option is IQ2_XXS at 9.8 GB.
VRAM requirement by quantization
★ Recommended — best balance of quality and VRAM usage.
Can I run Cosmos Reason2 32B on a Mac?
Yes — Mac Mini M2 Pro (16 GB) and 145 other Macs can run Cosmos Reason2 32B, the smallest at IQ2_XXS. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 20.6 GB of usable unified memory (RAM minus macOS overhead), from Mac Studio M1 Max (32 GB).
Can I run Cosmos Reason2 32B locally?
Yes — Cosmos Reason2 32B can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.
How fast is Cosmos Reason2 32B?
At Q4_K_M, Cosmos Reason2 32B can reach ~232 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~32 tok/s. 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 ÷ 20.6 × 0.65 = ~252 tok/s
Estimated speed at Q4_K_M (20.6 GB)
Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
What's the download size of Cosmos Reason2 32B?
At Q4_K_M, the download is about 20.01 GB. The full-precision BF16 version is 66.71 GB. The smallest option (IQ2_XXS) is 9.17 GB.
Which GPUs can run Cosmos Reason2 32B?
7 consumer GPUs can run Cosmos Reason2 32B at Q4_K_M. Top options include NVIDIA GeForce RTX 5090, AMD Radeon RX 7900 XTX, NVIDIA GeForce RTX 3090. 1 GPU has plenty of headroom for comfortable inference.
Which devices can run Cosmos Reason2 32B?
112 devices with unified memory can run Cosmos Reason2 32B at Q4_K_M (20.6 GB), including AMD Ryzen 7 8845HS (Hawk Point) Laptop, AMD Ryzen AI 7 350 (Krackan Point) Laptop, AMD Ryzen AI 9 HX 370 (Strix Point) Laptop, ASUS Ascent GX10. Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.