Can the RTX 3090 Run Qwen2.5 14B?
Runs
Yes. Qwen2.5 14B at Q4_K_M is a 9 GB download (Hugging Face file listing for bartowski/Qwen2.5-14B-Instruct-GGUF), and the RTX 3090 has 24 GB of VRAM, leaving about 15 GB for the KV cache and runtime. Published runs put generation at a median of 55.8 tokens per second across 3 cited measurements. Fastest cited run: LocalScore (Mozilla Builders).
Buy the RTX 3090
ZOTAC Gaming GeForce RTX 3090 Trinity OC 24GB GDDR6X 384-bit 19.5 Gbps PCIE 4.0 Gaming…Tracked at $1,900 on 2026-11-07 — price may vary.
Published runs for Qwen2.5 14B on the RTX 3090
| Generation | Quant | Runtime | Context | Source | Date |
|---|---|---|---|---|---|
| 56.1 tok/s | q4_K_M | llama.cpp | — | LocalScore (Mozilla Builders)labelled "qwen2.5:14b" | 2025-06-01 |
| 55.8 tok/s | q4_K_M | ollama | — | LocalScore (Mozilla Builders)labelled "qwen2.5:14b" | date not given |
| 55 tok/s | q4_K_M | llama.cpp | — | kunalganglani.com LLM Benchmarkslabelled "qwen2.5:14b" | 2024-12-01 |
More about this pair
How much VRAM does Qwen2.5 14B need?
The Q4_K_M file is 9 GB. Add roughly 1.5 GB for the runtime and an ~8K-token context, so about 10.5 GB of VRAM holds it comfortably. Longer contexts grow the KV cache and need more.
What speed does Qwen2.5 14B reach on the RTX 3090?
The fastest published run is 56.1 tokens per second (LocalScore (Mozilla Builders), q4_K_M, llama.cpp); the median of 3 cited runs is 55.8 tok/s. Generation reads every weight once per token, so the RTX 3090's 936 GB/s of memory bandwidth divided by the 8.99 GB file puts the theoretical ceiling near 104 tokens per second. That is an upper bound, not a speed estimate: kernel, launch and compute overhead keep real runs below it, and SpecPicks rejects any published figure above it.
Check another pair
Qwen2.5 14B on other GPUs
- RTX 4060 — Offload
- RTX 3060 Ti — Offload
- RTX 3060 12GB — Runs, 26.5 tok/s median
- Intel Arc B580 — Runs
- RTX 4070 SUPER — Runs, 24.9 tok/s (1 run)
- RTX 4060 Ti 16GB — Runs, 27.2 tok/s median
- RTX 5060 Ti 16GB — Runs, 32.8 tok/s median
- Intel Arc A770 16GB — Runs, 21.4 tok/s (1 run)
- Radeon RX 9070 XT — Runs
- RTX 4070 Ti SUPER — Runs, 36.8 tok/s (1 run)
- RTX 4080 — Runs
- RTX 5070 Ti — Runs, 55.1 tok/s median
- RTX 5080 — Runs, 85 tok/s median
- Tesla P40 24GB — Runs, 14.5 tok/s median
- Radeon RX 7900 XTX — Runs, 46.8 tok/s median
- RTX 4090 — Runs, 63.9 tok/s (1 run)
- RTX 5090 — Runs, 89.9 tok/s (1 run)
- RTX A6000 48GB — Runs, 50.3 tok/s (1 run)
- RTX PRO 5000 Blackwell 48GB — Runs, 54.7 tok/s (1 run)
- Instinct MI210 64GB — Runs
- RTX PRO 6000 Blackwell 96GB — Runs, 81.9 tok/s (1 run)
Other models on the RTX 3090
- Llama 3.2 1B — Runs, 267 tok/s (1 run)
- Llama 3.2 3B — Runs
- Mistral 7B — Runs
- Qwen2.5 7B — Runs
- Llama 3.1 8B — Runs, 92 tok/s median
- DeepSeek-R1-Distill-Llama 8B — Runs
- Qwen3 8B — Runs, 101 tok/s median
- Gemma 2 9B — Runs
- Gemma 3 12B — Runs
- DeepSeek-R1-Distill-Qwen 14B — Runs, 37.5 tok/s (1 run)
- Qwen3 14B — Runs, 61.1 tok/s median
- Phi-4 14B — Runs
- gpt-oss-20b — Runs
- Gemma 3 27B — Runs
- Gemma 2 27B — Runs
- Qwen3 30B-A3B — Runs
- Qwen3 32B — Runs, 32.7 tok/s median
- Qwen2.5 32B — Runs, 23 tok/s median
- DeepSeek-R1-Distill-Qwen 32B — Runs
- Qwen3 32B (Q6_K) — Offload
- Qwen2.5 32B (Q6_K) — Offload
- DeepSeek-R1-Distill-Qwen 32B (Q6_K) — Offload
- Gemma 3 27B (Q8_0) — Offload
- Llama 3.1 70B — Offload
- Llama 3.3 70B — Offload
- DeepSeek-R1-Distill-Llama 70B — Offload
How the checks work
The fit is the Q4_K_M file size against the card's VRAM, with 1.5 GB kept for the runtime and an ~8K context. A published run counts only if the pair fits, the run is Q4-class, its source does not call it an estimate, it is under the bandwidth ceiling (memory bandwidth ÷ GB read per token: the whole file for a dense model, only the active experts' weights for a mixture-of-experts model such as gpt-oss-20b or Qwen3 30B-A3B, which is why those can post figures above bandwidth ÷ file size), it is not slower than the same card on a model twice its size, a single run is not slower than the same card on a larger dense model, and it is not one figure pasted across many cards. The whole matrix is downloadable as a CSV under CC BY 4.0.
As an Amazon Associate, SpecPicks earns from qualifying purchases. As an eBay Partner Network member, SpecPicks may earn a commission on eBay links.
This page is editorial synthesis based on publicly available information. No independent first-party benchmarking is reported.