Can I run Qwen3 14B on the GeForce RTX 4060?
Runs slowlyQ4_K_M is 3.5 GB short of fitting in VRAM, so 39% of the weights run from system RAM: Runs slowly, est. 8.1 tok/s (6.5–9.7, calibrated estimate ±20%). Q2_K (heavy quality loss) would reach Runs well, est. 23.3 tok/s (18.7–28.0, calibrated estimate ±20%).
8 GB VRAM · 272.0 GB/s · FP16 15.1 TFLOPS. Assumes 32 GB of DDR5-5600 system RAM, Windows with this GPU driving the display, 8k context and an f16 KV cache.
Every quant of Qwen3 14B on the GeForce RTX 4060
| Quant | File size | Verdict | Speed | Memory | Runs as | Notes |
|---|---|---|---|---|---|---|
| Q2_K | 5.75 GB | Runs well | est. 23.3 tok/s18.7–28.0calibrated estimate ±20% | 8.0 / 8.0 GB + 0.3 GB RAM | Partial offload | Not fully on the GPU — Runs great needs the whole model in GPU memory |
| IQ4_XS | 8.14 GB | Runs slowly | est. 9.7 tok/s7.8–11.7calibrated estimate ±20% | 8.0 / 8.0 GB + 2.7 GB RAM | Partial offload | Less than 90% on the GPU — a GPU/CPU split is capped at Runs slowly |
| Q4_K_Mbaseline | 9.00 GB | Runs slowly | est. 8.1 tok/s6.5–9.7calibrated estimate ±20% | 8.0 / 8.0 GB + 3.5 GB RAM | Partial offload | Less than 90% on the GPU — a GPU/CPU split is capped at Runs slowly |
| Q8_0 | 15.70 GB | Runs slowly | est. 3.6 tok/s2.9–4.3calibrated estimate ±20% | 8.0 / 8.0 GB + 10.2 GB RAM | Partial offload | — |
Where the memory goes at Q4_K_M
- Weights
- 5.5 GB
- KV cache
- 1.3 GB
- Compute buffer
- 0.6 GB
- OS reserve
- 0.6 GB
- Weights in system RAM
- 3.5 GB
Context length vs. KV cache
| Context | KV f16 | KV q8_0 | KV q4_0 |
|---|---|---|---|
| 4k | 10.2 GBRuns slowly est. 9.8 tok/s7.8–11.7calibrated estimate ±20% | 9.9 GBRuns slowly est. 10.7 tok/s8.6–12.8calibrated estimate ±20% | 9.7 GBRuns slowly est. 11.3 tok/s9.0–13.5calibrated estimate ±20% |
| 8k | 10.9 GBRuns slowly est. 8.1 tok/s6.5–9.7calibrated estimate ±20% | 10.3 GBRuns slowly est. 9.6 tok/s7.7–11.5calibrated estimate ±20% | 10.0 GBRuns slowly est. 10.5 tok/s8.4–12.7calibrated estimate ±20% |
| 16k | 12.3 GBRuns slowly est. 5.9 tok/s4.7–7.0calibrated estimate ±20% | 11.1 GBRuns slowly est. 7.8 tok/s6.3–9.4calibrated estimate ±20% | 10.4 GBRuns slowly est. 9.3 tok/s7.5–11.2calibrated estimate ±20% |
| 32k | 15.2 GBRuns slowly est. 3.5 tok/s2.8–4.2calibrated estimate ±20% | 12.7 GBRuns slowly est. 5.5 tok/s4.4–6.6calibrated estimate ±20% | 11.3 GBRuns slowly est. 7.5 tok/s6.0–9.0calibrated estimate ±20% |
S = Runs great · A = Runs well · B = Runs slowly · F = Won't run
At Q4_K_M the verdict stays Runs slowly up to 32k context with an f16 KV cache, and up to 32k with q8_0.
Estimated speed
- Generation
- est. 8.1 tok/s (6.5–9.7, calibrated estimate ±20%)
- Prompt processing
- Prompt-processing estimates only apply when the whole model runs on the GPU.
Why this verdict: Less than 90% on the GPU — a GPU/CPU split is capped at Runs slowly
Measured on this exact combination
No public measurement for this exact combination yet; the numbers above are estimates.
If this is not enough
A smaller model that runs well on the GeForce RTX 4060
Runs greatQwen3.5 9B: Runs great, est. 31.0 tok/s (27.3–34.7, calibrated estimate ±12%)
A lower quant of Qwen3 14B
Runs wellQ2_K (5.75 GB, heavy quality loss): Runs well, est. 23.3 tok/s (18.7–28.0, calibrated estimate ±20%)
Cheapest GPU that runs Qwen3 14B great
Runs greatGeForce RTX 3060 12GB — $250 street (as of 2026-08-09): est. 24.4 tok/s (21.4–27.3, calibrated estimate ±12%).
How to run it
Commands for the Q4_K_M file. Both tools download from Hugging Face on first run.
File: Qwen3-14B-Q4_K_M.gguf (9.00 GB) from unsloth/Qwen3-14B-GGUF.
llama-server -hf unsloth/Qwen3-14B-GGUF:Q4_K_M -c 8192 -ngl 24 -fa on- -ngl 24 puts 24 of 40 layers on the GPU; if VRAM overflows, lower it by 1–2.
- -fa on enables flash attention (needed for KV cache quantization).
- --cache-type-k q8_0 --cache-type-v q8_0 shrinks the KV cache to 0.72 GB at 8k context.
PowerShell (quit the Ollama tray app first):
$env:OLLAMA_CONTEXT_LENGTH="8192"; ollama serve
ollama run hf.co/unsloth/Qwen3-14B-GGUF:Q4_K_M- Ollama defaults to 4k context on GPUs under 24 GB. Set OLLAMA_CONTEXT_LENGTH when starting the server, or /set parameter num_ctx inside the chat.
- Ollama splits layers between GPU and CPU automatically.
/set parameter num_ctx 8192
Related pages
Guide: How much VRAM do local LLMs need? 8–32 GB tiers (2026)
Other models on the GeForce RTX 4060
Qwen3 14B on other hardware
Frequently asked questions
Can I run Qwen3 14B on the GeForce RTX 4060?
Q4_K_M is 3.5 GB short of fitting in VRAM, so 39% of the weights run from system RAM: Runs slowly, est. 8.1 tok/s (6.5–9.7, calibrated estimate ±20%). Q2_K (heavy quality loss) would reach Runs well, est. 23.3 tok/s (18.7–28.0, calibrated estimate ±20%). Q4_K_M needs 10.9 GB at 8k context; this setup has 7.4 GB of usable memory and 28.0 GB of free system RAM.
How much context can Qwen3 14B use on the GeForce RTX 4060?
At Q4_K_M the verdict stays Runs slowly up to 32k context with an f16 KV cache (5.4 GB of KV) and up to 32k with a q8_0 KV cache (2.9 GB). The model supports up to 32k.
Which quant should I use, and how fast is it?
Q4_K_M (9.00 GB) is the recommended balance: Runs slowly, est. 8.1 tok/s (6.5–9.7, calibrated estimate ±20%). The largest tracked file that still gets Runs slowly is Q8_0 (15.70 GB), est. 3.6 tok/s (2.9–4.3, calibrated estimate ±20%).
Speeds are estimates from memory bandwidth, calibrated against public benchmarks, and each one comes with an error band and a confidence label. Real results vary with drivers, backend, context length and thermals.