NVIDIA RTX 4500 Ada
The NVIDIA RTX 4500 Ada has 24 GB VRAM and 432 GB/s memory bandwidth. It can run 52 of our 94 tracked models natively in VRAM at 8k context.
With 24 GB GDDR6, the NVIDIA RTX 4500 Ada is a workstation-tier GPU that can run 52 models natively. It comfortably runs 7B–32B models at Q4; 70B-class models typically need CPU offload.
The NVIDIA RTX 4500 Ada is the mid-range Ada Lovelace workstation GPU with 24GB ECC GDDR6 on a 192-bit bus at 432 GB/s. With 7,680 CUDA cores it balances professional reliability with meaningful LLM inference headroom, fitting 13B models at Q8_0 and 22B–27B models at Q4_K_M entirely in VRAM. The dual-slot blower cooler allows dense multi-card workstation configurations without sacrificing airflow.
NVIDIA RTX 4500 Ada: 2023 Ada Lovelace mid-range workstation card: 24GB ECC GDDR6 on a 192-bit bus at 432 GB/s, dual-slot blower cooler.
13B models fit at Q8_0; 22B-27B models fit at Q4_K_M entirely in VRAM. ~18-25 t/s for 7B Q4.
Full CUDA with ECC. The blower cooler design allows dense multi-card configurations in a single workstation chassis.
| Vendor | NVIDIA |
| Architecture | Ada Lovelace |
| VRAM | 24 GB |
| Memory type | GDDR6 |
| Memory bandwidth | 432 GB/s |
| Compute backend | CUDA |
| Tier | Workstation |
| Released | 2023 |
| Models (native) | 52 / 94 |
| Models (offload) | 7 / 94 |
Software: Full llama.cpp and Ollama support out of the box. CUDA 12.x recommended; driver ≥ 525 required.
Popular models for this GPU
Models this GPU runs natively in VRAM (52)
- Mixtral 8x7B Instruct v0.146.7B · MMLU-Pro 29.7Q2_K · ~16.1 t/s
- Qwen 3.5 35B-A3B (MoE)35B · MMLU-Pro 85.3Q3_K_M · ~56.4 t/s
- Qwen 3.6 35B35B · MMLU-Pro 85.2Q3_K_M · ~14.8 t/s
- Ornith 1.5 35B-A3B (MoE)35B · MMLU-Pro N/AQ3_K_M · ~56.4 t/s
- Yi 1.5 34B Chat34.4B · MMLU-Pro 37.0Q3_K_M · ~15.1 t/s
Show 47 more
- Qwen3 32B32.8B · MMLU-Pro 65.5Q3_K_M · ~16.4 t/s
- Qwen 2.5 32B Instruct32.5B · MMLU-Pro 69.0Q3_K_M · ~15.8 t/s
- Qwen 2.5 Coder 32B Instruct32.5B · MMLU-Pro 62.3Q3_K_M · ~15.8 t/s
- DeepSeek R1 Distill Qwen 32B32.5B · MMLU-Pro 65.0Q3_K_M · ~15.8 t/s
- Nemotron 3 Nano 30B32B · MMLU-Pro 78.3Q4_K_M · ~43 t/s
- Gemma 4 31B30.7B · MMLU-Pro 85.2Q4_K_M · ~13.9 t/s
- Qwen3 30B-A3B (MoE)30B · MMLU-Pro 61.5Q4_K_M · ~40.7 t/s
- Nemotron 3.5 Lightning 30B-A3B30B · MMLU-Pro 81.6Q4_K_M · ~45.7 t/s
- Muse Glimmer 30B27.8B · MMLU-Pro N/AQ5_K_M · ~14.1 t/s
- Gemma 2 27B Instruct27.2B · MMLU-Pro 38.0Q4_K_M · ~14.3 t/s
- Gemma 3 27B Instruct27B · MMLU-Pro 67.5Q4_K_M · ~15.6 t/s
- Qwen 3.6 27B27B · MMLU-Pro 86.2Q5_K_M · ~14.2 t/s
- UI-Mate 27B27B · MMLU-Pro ~86.2Q5_K_M · ~14.2 t/s
- Bonsai 27B27B · MMLU-Pro 81.5Ternary (Q2_0) · ~31.9 t/s
- Qwen 3.8 27B27B · MMLU-Pro N/AQ5_K_M · ~14.2 t/s
- Gemma 4 26B (MoE)25.2B · MMLU-Pro 82.6Q5_K_M · ~29.9 t/s
- Mistral Small 3.1 24B Instruct24B · MMLU-Pro 66.8Q5_K_M · ~15.2 t/s
- Mistral Small 22B22.2B · MMLU-Pro 49.2Q6_K · ~14 t/s
- GPT-OSS 20B21B · MMLU-Pro 67.9Q6_K · ~27.9 t/s
- Qwen3 14B14.8B · MMLU-Pro 61.0Q8_0 · ~16.4 t/s
- Qwen 2.5 14B Instruct14.7B · MMLU-Pro 63.7Q8_0 · ~16.3 t/s
- Phi-4 14B Instruct14B · MMLU-Pro 70.4Q8_0 · ~17.3 t/s
- Mistral Nemo 12B Instruct12.2B · MMLU-Pro 35.6Q8_0 · ~19.6 t/s
- Gemma 3 12B Instruct12.2B · MMLU-Pro 60.6Q8_0 · ~20 t/s
- Gemma 4 12B (Unified)12B · MMLU-Pro 77.2Q8_0 · ~17.6 t/s
- Gemma 2 9B Instruct9.2B · MMLU-Pro 32.0Q8_0 · ~22.3 t/s
- Qwen 3.5 9B9B · MMLU-Pro 82.5BF16 · ~15.4 t/s
- Ornith 1.5 9B9B · MMLU-Pro N/ABF16 · ~15.4 t/s
- Llama 3.1 8B Instruct8B · MMLU-Pro 48.3BF16 · ~16.4 t/s
- DeepSeek R1 Distill Llama 8B8B · MMLU-Pro 41.0BF16 · ~16.4 t/s
- Qwen3 8B8B · MMLU-Pro 56.7BF16 · ~16.3 t/s
- Qwen 2.5 7B Instruct7.6B · MMLU-Pro 56.3BF16 · ~17.9 t/s
- Mistral 7B Instruct v0.37.25B · MMLU-Pro 30.0BF16 · ~18 t/s
- Gemma 3 4B Instruct4B · MMLU-Pro 43.6FP32 · ~17 t/s
- Gemma 4 E4B4B · MMLU-Pro 69.4FP32 · ~16.5 t/s
- Phi-3.5 Mini Instruct3.8B · MMLU-Pro 47.4FP32 · ~15.2 t/s
- Phi-4-mini Instruct3.8B · MMLU-Pro 67.3FP32 · ~17.3 t/s
- Llama 3.2 3B Instruct3.2B · MMLU-Pro 24.0FP32 · ~20.4 t/s
- Qwen 2.5 3B Instruct3.1B · MMLU-Pro 32.4FP32 · ~22.1 t/s
- Gemma 2 2B Instruct2.6B · MMLU-Pro 17.8FP32 · ~24.9 t/s
- Gemma 4 E2B2B · MMLU-Pro 60.0FP32 · ~33.4 t/s
- SmolLM2 1.7B Instruct1.7B · MMLU-Pro 19.0FP32 · ~33.4 t/s
- Qwen 2.5 1.5B Instruct1.5B · MMLU-Pro 16.8FP32 · ~45 t/s
- Llama 3.2 1B Instruct1.24B · MMLU-Pro 12.5FP32 · ~53.7 t/s
- Gemma 3 1B Instruct1B · MMLU-Pro 14.7FP32 · ~64.9 t/s
- Qwen 2.5 0.5B Instruct0.5B · MMLU-Pro 10.0FP32 · ~133.7 t/s
- SmolLM2 360M Instruct0.36B · MMLU-Pro 8.0FP32 · ~158.1 t/s
Models that fit with CPU offload (7)
These use system RAM for layers that don't fit in VRAM, so expect much slower inference.
- GLM-4.5 Air 106B106B · MMLU-Pro 81.4Q2_K · ~3 t/s
- GLM-4.6V 106B106B · MMLU-Pro 79.9Q2_K · ~3 t/s
- Qwen 2.5 72B Instruct72B · MMLU-Pro 71.1Q3_K_M · ~1.5 t/s
- Llama 3.3 70B Instruct70B · MMLU-Pro 68.9Q3_K_M · ~1.6 t/s
- DeepSeek R1 Distill Llama 70B70B · MMLU-Pro 70.0Q3_K_M · ~1.6 t/s
- Llama 3.1 70B Instruct70B · MMLU-Pro 66.4Q3_K_M · ~1.6 t/s
- Command-R 35B35B · MMLU-Pro 33.0Q6_K · ~1.3 t/s
Too large for this GPU (35)
- Mixtral 8x22B Instruct v0.1
- Llama 3.1 405B Instruct
- DeepSeek V3 671B
- DeepSeek R1 671B
- Llama 4 Scout 109B
- Llama 4 Maverick 400B
- Qwen3 235B-A22B (MoE)
- MiniMax M1 456B
- GPT-OSS 120B
- GLM-4.5 355B
- GLM-4.6 355B
- GLM-4.7 358B
- Qwen 3.5 122B-A10B (MoE)
- MiniMax M2.5 229B
- GLM-5 744B
- MiniMax M2.7 229B
- Nemotron 3 Super 120B
- Kimi K2.6
- GLM-5.1 754B
- DeepSeek V4 Pro 1.6T
- DeepSeek V4 Flash 284B
- Mistral Medium 3.5 128B
- GLM-5.2 753B
- Nemotron 3 Ultra 550B-A55B
- Step 3.5 Flash
- Step 3.7 Flash
- MiMo V2.5 Pro
- Kimi K2.5
- MiniMax M3
- Inkling
- Kimi K3
- DeepSeek V4 Flash 0731 284B
- Qwen3.8 2.4T-A95B
- DeepSeek V4 Pro 0813 1.6T
- Ornith 1.5 397B (MoE)
Frequently asked questions
- How much VRAM does the NVIDIA RTX 4500 Ada have?
- The NVIDIA RTX 4500 Ada has 24 GB of GDDR6 with 432 GB/s memory bandwidth.
- What is the NVIDIA RTX 4500 Ada best for?
- With 24 GB of VRAM, the NVIDIA RTX 4500 Ada is well-suited for running 7B–32B models at Q4 with room for context, making it a great all-rounder for local LLM inference.
- What LLMs can the NVIDIA RTX 4500 Ada run locally?
- The NVIDIA RTX 4500 Ada can run 52 of the 94 open-weight models tracked by CanItRun natively in VRAM at 8k context. Top options include: Qwen 3.8 27B at Q5_K_M, Muse Glimmer 30B at Q5_K_M, Ornith 1.5 9B at BF16.
- Can the NVIDIA RTX 4500 Ada run Gemma 4 31B?
- Yes. The NVIDIA RTX 4500 Ada runs Gemma 4 31B natively in VRAM at Q4_K_M quantization, achieving approximately 13.9 tokens per second.
- Can the NVIDIA RTX 4500 Ada run Qwen 3.6 27B?
- Yes. The NVIDIA RTX 4500 Ada runs Qwen 3.6 27B natively in VRAM at Q5_K_M quantization, achieving approximately 14.2 tokens per second.
- Can the NVIDIA RTX 4500 Ada run Qwen3 8B?
- Yes. The NVIDIA RTX 4500 Ada runs Qwen3 8B natively in VRAM at BF16 quantization, achieving approximately 16.3 tokens per second.