EXAONE 4.5 33B — Hardware Requirements & GPU Compatibility
VisionFunctionsEXAONE 4.5 stands as LG AI Research's first open-weight vision-language release, boasting a total of 33 billion parameters. This system integrates a dedicated 1.29-billion-parameter vision encoder with the EXAONE 4.0 language model (31.7 billion parameters), utilizing a hybrid attention design that combines sliding-window and global layers. According to its card, it beats state-of-the-art models of comparable size when handling document understanding and Korean contextual reasoning, while supporting English, Korean, Spanish, German, Japanese, and Vietnamese. To run it locally, you need at least 24 GB of GPU memory after quantization or a machine with unified memory. The model handles a context window of up to 262,144 tokens. It comes under the EXAONE AI Model License Agreement 1.2 - NC, which is non-commercial, meaning commercial use isn't allowed per its terms. Published in April 2026, it builds on prior EXAONE 4.0 language models.
Decent on a 24 GB GPU
Runs on 24 GB (Q4_K_M · ~31 tok/s · 8K)Macs from 24 GB (IQ3_M · ~13 tok/s)
Other
Reference GPU · RTX 4090 24 GB
Estimates, ±20% · how we estimate
Run EXAONE 4.5 33B
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/LGAI-EXAONE/EXAONE-4.5-33B-GGUF:Q4_K_MQ4_K_M · needs ~25.2 GB · 16K context
Download: EXAONE-4.5-33B-Q4_K_M.gguf20.05 GBLGAI-EXAONE
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) (6 GB headroom) · Q4_K_M
- Generation speed
- ~54 tok/s
- generation speed
- Cost per 1M output tokens
- $2.75
- per 1M output tokens
How Much VRAM Does EXAONE 4.5 33B Need?
Select a quantization level; the listed hardware adheres to it.
| Quantization | VRAM | +32K | On RTX 4090 (reference) | |
|---|---|---|---|---|
| 15.4 GB | 23.5 GB | 2-bit · K-quant — 2-bit quantization with K-quant improvements | Fits: 32K context · ~42 tok/s | |
| 15.9 GB | 23.9 GB | 3-bit small — 3-bit small quantization | Fits: 32K context · ~41 tok/s | |
| 17.6 GB | 25.6 GB | 3-bit medium — 3-bit medium quantization | Fits: 16K context · ~37 tok/s | |
| 18.0 GB | 26.1 GB | 4-bit · legacy — 4-bit legacy quantization | Fits: 16K context · ~36 tok/s | |
| 21.4 GB | 29.5 GB | 4-bit medium — 4-bit medium quantization — most popular sweet spot | Fits: 8K context · ~31 tok/s | |
| 25.3 GB | 33.4 GB | 5-bit medium — 5-bit medium quantization — good quality/size tradeoff | Does not fit: needs 29.1 GB | |
| 29.2 GB | 37.2 GB | 6-bit, very good — 6-bit quantization, very good quality | Does not fit: needs 32.9 GB | |
| 35.2 GB | 43.2 GB | 8-bit, near-lossless — 8-bit quantization, near-lossless | Does not fit: needs 39 GB |
Selected Q4_K_M
4-bit medium quantization25.2GBat 16K context
- Weights
- 20.6 GB
- KV cache
- 4.3 GB
- Overhead
- 0.3 GB
Fits RTX 4090 (reference) · fits up to 8K context · ~31 tok/s
≥50 tok/s needs ≥1.65 TB/s · RTX 5090 ~54
Or .
Which Hardware Can Run EXAONE 4.5 33B?
Q4_K_M · 21.4 GBEXAONE 4.5 33B (Q4_K_M) requires 21.4 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 28+ GB is recommended. Using the full 262K context window can add up to 68.2 GB, bringing total usage to 89.6 GB. 7 GPUs can run it, including NVIDIA GeForce RTX 5090. 112 devices with unified memory can run EXAONE 4.5 33B, 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 Q3_K_S (19.6 GB)
1 GPU
RX 7900 XT~30Simulate output on RX 7900 XT32–40 GB
fits · 6.8 GB free · up to 32K ctx
1 GPU · 3 Mac classes · 9 device families
RTX 5090~54Simulate output on RTX 5090M5 Max*36 GB~15Simulate output on Mac Studio M5 Max (36 GB)M632 GB~6Simulate output on iMac 24" M6 (32 GB)RTX Spark N1X~9Simulate 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
Alternatives to EXAONE 4.5 33B
EXAONE family · 4 models →Similar models
6| Model | Score | VRAM | tok/s | Why it is similar |
|---|---|---|---|---|
| Nanbeige4.3 33BNanbeige | — | 23.7 GB | ~32 | same size |
| Sage A1.5InternScience | — | 22 GB | ~151 | same size |
| MiniCPM V 4.7 35B A3BOpenBMB | — | 22.1 GB | ~169 | same size |
| Llm Jp 4.1 32B A3b Thinkingllm-jp | — | 20.1 GB | ~188 | same size |
| Llm Jp 4.1 33B Thinkingllm-jp | — | 22.5 GB | ~35 | same size |
| Agnes 3.0 FlashAgnes-AI | — | 22.1 GB | ~35 | same size |
Specifications
- Publisher
- LGAI-EXAONE
- Family
- EXAONE
- Parameters
- 34.4B
- Architecture
- Exaone4_5_ForConditionalGeneration
- Context Length
- 262,144 tokens
- Vocabulary Size
- 153,600
- Release Date
- 2026-04-04
- License
- Other
Frequently Asked Questions
How much VRAM does EXAONE 4.5 33B need?
EXAONE 4.5 33B requires 21.4 GB of VRAM at Q4_K_M, or 69.5 GB at BF16. Full 262K context adds up to 68.2 GB (89.6 GB total).
VRAM = Weights + KV Cache + Overhead
Weights = 34.4B × 4.8 bits ÷ 8 = 20.6 GB
KV Cache + Overhead ≈ 0.8 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 ≈ 69 GB (at full 262K context)
VRAM usage by quantization
Can NVIDIA GeForce RTX 4090 run EXAONE 4.5 33B?
Yes, at Q4_K_M (21.4 GB) or lower. Higher quantizations like Q5_K_S (24.4 GB) exceed the NVIDIA GeForce RTX 4090's 24 GB.
What's the best quantization for EXAONE 4.5 33B?
For EXAONE 4.5 33B, Q4_K_M (21.4 GB) offers the best balance of quality and VRAM usage. Q5_K_S (24.4 GB) provides better quality if you have the VRAM. The smallest option is IQ3_S at 15.4 GB.
VRAM requirement by quantization
★ Recommended — best balance of quality and VRAM usage.
Can I run EXAONE 4.5 33B on a Mac?
Yes — Mac Mini M5 Pro (24 GB) and 115 other Macs can run EXAONE 4.5 33B, the smallest at IQ3_M. Apple Silicon uses unified memory, so the model shares RAM with the system. At Q4_K_M you need at least 21.4 GB of usable unified memory (RAM minus macOS overhead), from Mac Studio M1 Max (32 GB).
Can I run EXAONE 4.5 33B locally?
Yes — EXAONE 4.5 33B can run locally on consumer hardware. Popular tools include Ollama, LM Studio, and llama.cpp.
How fast is EXAONE 4.5 33B?
At Q4_K_M, EXAONE 4.5 33B can reach ~224 tok/s on AMD Instinct MI350X. On NVIDIA GeForce RTX 4090: ~31 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 ÷ 21.45 × 0.65 = ~242 tok/s
Estimated speed at Q4_K_M (21.4 GB)
Real-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
What's the download size of EXAONE 4.5 33B?
At Q4_K_M, the download is about 20.61 GB. The full-precision BF16 version is 68.70 GB. The smallest option (IQ3_S) is 14.60 GB.
Which GPUs can run EXAONE 4.5 33B?
7 consumer GPUs can run EXAONE 4.5 33B at Q4_K_M. Top options include AMD Radeon RX 7900 XTX, NVIDIA GeForce RTX 3090.
Which devices can run EXAONE 4.5 33B?
112 devices with unified memory can run EXAONE 4.5 33B at Q4_K_M (21.4 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.