NVIDIA · Ampere · 24 GB VRAM tier
Best AI Models for NVIDIA RTX A5000 (24 GB)
Runs 27–31B and 35B MoE models well; 70B doesn't fit in 24 GB.
| # | Model | Speed | llmrun Score | Quant · VRAM |
|---|---|---|---|---|
| 1 | Qwen3.8 27BAlibabaHow to run → Q4_K_M · 17.9 GB · 28.7 t/s | 28.7 tok/s | Q4_K_M · 17.9 GB | |
| 2 | Qwen3.6 27BAlibaba Q4_K_M · 17.9 GB · 28.7 t/s | 28.7 tok/s | Q4_K_M · 17.9 GB | |
| 3 | Gemma 4 31B ITGoogle Q4_K_M · 21.7 GB · 24.5 t/s | 24.5 tok/s | Q4_K_M · 21.7 GB | |
| 4 | Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB · 140 t/s | 140 tok/s | Q4_K_M · 16.5 GB | |
| 5 | Q4_K_M · 19.3 GB · 174 t/s | 174 tok/s | Q4_K_M · 19.3 GB |
| # | Model | Speed | Coding Score | Quant · VRAM |
|---|---|---|---|---|
| 1 | Qwen3.8 27BAlibabaHow to run → Q4_K_M · 17.9 GB · 28.7 t/s | 28.7 tok/s | Q4_K_M · 17.9 GB | |
| 2 | Gemma 4 31B ITGoogle Q4_K_M · 21.7 GB · 24.5 t/s | 24.5 tok/s | Q4_K_M · 21.7 GB | |
| 3 | Qwen3.6 35B A3BAlibaba Q4_K_M · 22.0 GB · 125 t/s | 125 tok/s | Q4_K_M · 22.0 GB | |
| 4 | Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB · 140 t/s | 140 tok/s | Q4_K_M · 16.5 GB | |
| 5 | Q4_K_M · 19.3 GB · 174 t/s | 174 tok/s | Q4_K_M · 19.3 GB |
| # | Model | Speed | Reasoning Score | Quant · VRAM |
|---|---|---|---|---|
| 1 | Qwen3.8 27BAlibabaHow to run → Q4_K_M · 17.9 GB · 28.7 t/s | 28.7 tok/s | Q4_K_M · 17.9 GB | |
| 2 | Gemma 4 31B ITGoogle Q4_K_M · 21.7 GB · 24.5 t/s | 24.5 tok/s | Q4_K_M · 21.7 GB | |
| 3 | Qwen3.5 27BAlibaba Q4_K_M · 17.9 GB · 28.7 t/s | 28.7 tok/s | Q4_K_M · 17.9 GB | |
| 4 | Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB · 140 t/s | 140 tok/s | Q4_K_M · 16.5 GB | |
| 5 | Q4_K_M · 19.3 GB · 174 t/s | 174 tok/s | Q4_K_M · 19.3 GB |
| # | Model | Speed | Math Score | Quant · VRAM |
|---|---|---|---|---|
| 1 | Qwen3.8 27BAlibabaHow to run → Q4_K_M · 17.9 GB · 28.7 t/s | 28.7 tok/s | Q4_K_M · 17.9 GB | |
| 2 | Qwen3.6 27BAlibaba Q4_K_M · 17.9 GB · 28.7 t/s | 28.7 tok/s | Q4_K_M · 17.9 GB | |
| 3 | Qwen3.6 35B A3BAlibaba Q4_K_M · 22.0 GB · 125 t/s | 125 tok/s | Q4_K_M · 22.0 GB | |
| 4 | Gemma 4 31B ITGoogle Q4_K_M · 21.7 GB · 24.5 t/s | 24.5 tok/s | overall | Q4_K_M · 21.7 GB |
| 5 | Q4_K_M · 19.3 GB · 174 t/s | 174 tok/s | Q4_K_M · 19.3 GB |
| # | Model | Speed | Long context Score | Quant · VRAM |
|---|---|---|---|---|
| 1 | Qwen3.8 27BAlibabaHow to run → Q4_K_M · 17.9 GB · 28.7 t/s | 28.7 tok/s | Q4_K_M · 17.9 GB | |
| 2 | Qwen3.6 27BAlibaba Q4_K_M · 17.9 GB · 28.7 t/s | 28.7 tok/s | Q4_K_M · 17.9 GB | |
| 3 | Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB · 140 t/s | 140 tok/s | Q4_K_M · 16.5 GB | |
| 4 | GPT OSS 20BOpenAI Q4_K_M · 13.1 GB · 154 t/s | 154 tok/s | Q4_K_M · 13.1 GB | |
| 5 | LFM2.5 24B A2BLiquid AI Q4_K_M · 15.0 GB · 191 t/s | 191 tok/s | Q4_K_M · 15.0 GB |
| # | Model | Speed ↓ | llmrun Score | Quant · VRAM |
|---|---|---|---|---|
| 1 | GPT OSS 20BOpenAIHow to run → Q4_K_M · 13.1 GB · 154 t/s | 154 tok/s | Q4_K_M · 13.1 GB | |
| 2 | Q4_K_M · 19.3 GB · 174 t/s | 174 tok/s | Q4_K_M · 19.3 GB | |
| 3 | Qwen3.5 4BAlibaba Q4_K_M · 3.4 GB · 153 t/s | 153 tok/s | Q4_K_M · 3.4 GB | |
| 4 | Qwen3 30B A3B Thinking 2507Alibaba Q4_K_M · 19.4 GB · 148 t/s | 148 tok/s | Q4_K_M · 19.4 GB | |
| 5 | Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB · 140 t/s | 140 tok/s | Q4_K_M · 16.5 GB |
| # | Model | Speed | Vision Score | Quant · VRAM |
|---|---|---|---|---|
| 1 | Qwen3.8 27BAlibabaHow to run → Q4_K_M · 17.9 GB · 28.7 t/s | 28.7 tok/s | Q4_K_M · 17.9 GB | |
| 2 | Qwen3.6 27BAlibaba Q4_K_M · 17.9 GB · 28.7 t/s | 28.7 tok/s | Q4_K_M · 17.9 GB | |
| 3 | Gemma 4 31B ITGoogle Q4_K_M · 21.7 GB · 24.5 t/s | 24.5 tok/s | Q4_K_M · 21.7 GB | |
| 4 | Gemma 4 26B A4B ITGoogle Q4_K_M · 16.5 GB · 140 t/s | 140 tok/s | Q4_K_M · 16.5 GB | |
| 5 | JEV 27B VLautotrust Q4_K_M · 17.9 GB · 28.7 t/s | 28.7 tok/s | Q4_K_M · 17.9 GB |
The RTX A5000 runs 27–31B and 35B MoE models well. Ampere generation, released 2021.
RTX A5000 strengths and weaknesses
Strengths
- Runs 35B MoE models: Qwen3.6 35B A3B at 125 tok/s in 22 GB
- Room for long context: Qwen3.8 27B at Q4 uses 17.9 GB, leaving 6.1 GB free
Weaknesses
- 70B models don't fit: Llama 3.3 70B Instruct needs 51.3 GB at Q4 vs 24 GB here
- No FP8 or NVFP4: Ampere predates both; use GGUF, AWQ or GPTQ builds
Output speed depends on memory bandwidth (768 GB/s). Grey rows need more than 24 GB.
- 9BQwen3.5 9B · Q4_K_M · 6.6 GB78.5tok/s
- 14BQwen3 14B · Q4_K_M · 11.9 GB52.5tok/s
- 27–31BQwen3.8 27B · Q4_K_M · 17.9 GB28.7tok/s
- 35B MoEQwen3.6 35B A3B · Q4_K_M · 22.0 GB125tok/s
- 70BLlama 3.3 70B Instruct · not at 4-bitneeds 51.3 GB
- 120B MoEGPT OSS 120B · not at 4-bitneeds 70.8 GB
Estimated speeds: Q4_K_M, 4K context, one user. How we estimate
All 1141 models on the RTX A5000
Featured onlyBest quant that fits, ordered by llmrun Score. Fit grades how well it runs here, S best. How ranking works
| Model | Quant | VRAM | Speed | Context | Score | Fit |
|---|---|---|---|---|---|---|
| Q4_K_M | 11.5 GB | 58.2 t/s | — | A83, Runs well | ||
| Q4_K_M | 6.2 GB | 92.6 t/s | 8K | S90, Runs great | ||
Qwen2.5 Math 1.5BChatMathFunctions | Q4_K_M | 1.3 GB | 390.0 t/s | 4K | S100, Runs great | |
Qwen 1 8BChat | Q4_K_M | 1.9 GB | 412.6 t/s | 8K | S100, Runs great | |
SmolLM2 1.7BChat | Q4_K_M | 2.9 GB | 288.6 t/s | 8K | S100, Runs great | |
Llama 2 7B HFChat | Q4_K_M | 6.5 GB | 92.1 t/s | 4K | S90, Runs great | |
Nemotron H 8B Reasoning 128KChatReasoning | BF16 | 20.4 GB | 28.0 t/s | — | B62, Decent | |
Qwen3 Coder NextChatCodeFunctions | IQ2_XXS | 22.4 GB | 207.7 t/s | 262K | B62, Decent | |
| Q4_K_M | 0.1 GB | 5546.7 t/s | 1K | S100, Runs great | ||
Qwen1.5 7BChat | Q4_K_M | 13.5 GB | 83.1 t/s | 33K | S88, Runs great | |
DeepSeek Coder v2 Lite BaseChatCode | Q4_K_M | 10.2 GB | 202.4 t/s | 164K | S100, Runs great | |
Codegemma 7B ITChatCode | Q4_K_M | 8.2 GB | 88.7 t/s | — | S89, Runs great | |
Gemma 2B ITChat | Q4_K_M | 3.5 GB | 302.5 t/s | — | S100, Runs great | |
Internlm 7BChat | FP16 | 15.4 GB | 32.4 t/s | 2K | A72, Runs well | |
| Q4_K_M | 0.2 GB | 49920.0 t/s | — | S100, Runs great |
Rent NVIDIA RTX A5000 in the cloud
Hourly rate on demand for one RTX A5000.
Vast.aiMarketplace
from$0.23/hr
Buying one ($2,250) costs as much as ~9,800 hours of rental, about 41 months at 8 hours a day.
Cost per million tokens
Output-token cost on an RTX A5000 rented at $0.23/hr, one request at a time. Batch several users on the same card and the cost per token drops several-fold.
| Model | Quant | Speed | Per 1M output |
|---|---|---|---|
| NVIDIA Nemotron 3.5 Lightning 30B A3B BF16 | 174 tok/s | $0.37$0.0060 input | |
| Gemma 4 26B A4B IT | 140 tok/s | $0.46$0.011 input | |
| GLM 4.7 Flash | 118 tok/s | $0.54$0.012 input | |
| Qwen3.8 27B | 29 tok/s | $2.23$0.071 input | |
| Muse Glimmer 30B | 27 tok/s | $2.35$0.076 input |
Specs
- Architecture
- Ampere
- VRAM
- 24 GB GDDR6
- Memory bandwidth
- 768.0 GB/s
- FP16 (dense)
- 111.10 TFLOPS
- CUDA cores
- 8,192
- Tensor cores
- 256
- Compute capability
- 8.6
- TDP
- 230 W
- Released
- 2021-04-12
- MSRP
- $2,250
RTX A5000 FAQ
What is the largest model the RTX A5000 can run?
Qwen3.6 35B A3B (36.0B) is the largest at 4-bit or better: at Q4_K_M it uses 22.0 GB of 24 GB, and runs at about 125 tok/s.
Can the RTX A5000 run Qwen3.8 27B?
Yes. Qwen3.8 27B fits at Q4_K_M, using 17.9 GB of 24 GB, and it is the best model that runs well on this card.
Can the RTX A5000 run 70B models?
No. Llama 3.3 70B Instruct needs 51.3 GB at Q4, more than the 24 GB on this card.
How many tokens per second does the RTX A5000 give?
About 28.7 tok/s on Qwen3.8 27B at Q4_K_M, with the first token after 4.6 s on a 4K-token prompt.
What power supply does the RTX A5000 need?
Plan a 500 W PSU (2× the 230 W TDP).
Is 24 GB enough for coding and agents?
Yes for coding: Qwen3.8 27B, the best of the 9 coding models that fit, runs at about 29 tok/s. For agents, not yet: no agent-scored model fits in less than 96 GB.