NVIDIA RTX 3090 vs NVIDIA RTX 5060 Ti 16GB for local LLMs
226 reports on the NVIDIA RTX 3090 and 72 on the NVIDIA RTX 5060 Ti 16GB
Which is faster for local LLMs?
On Qwen3.8 27B at 3-bit, the NVIDIA RTX 3090 has 42.3 generation tokens/s (one run); prompt processing 1127.0 tokens/s (one run); the NVIDIA RTX 5060 Ti 16GB has 28.8 generation tokens/s (2 runs); prompt processing 400.0 tokens/s (one run). Those are community reports whose engines and settings differ, so the gap is a rough sign, not a controlled measure of the two cards. Memory bandwidth, which caps how fast a card can write, is 2.1x higher on the NVIDIA RTX 3090 (936 GB/s on the NVIDIA RTX 3090, 448 on the NVIDIA RTX 5060 Ti 16GB). Across 3 shared models with generation measurements on both cards, the NVIDIA RTX 3090's median is higher on 3 and the NVIDIA RTX 5060 Ti 16GB's on 0.
- Memory
- 24GB
- Memory bandwidth
- 936 GB/s
- FP16 compute
- 142 TFLOPS
- Reports
- 226
- Memory
- 16GB
- Memory bandwidth
- 448 GB/s
- FP16 compute
- 95 TFLOPS
- Reports
- 72
Measured on both
Generation and prompt processing speeds from plain runs (one device, one request, no speculative decoding, the whole model in the device's memory) of the same model size at the same bit class. Each side shows the engines and reported context behind its median. Context can mean a configured window or prompt depth, so these are community comparisons with differing settings. How to read these
| Model | NVIDIA RTX 3090 | NVIDIA RTX 5060 Ti 16GB |
|---|---|---|
| Qwen3.8 27B3-bit | 42.3one runPP 1127.0 (one run)llama.cpp16K context | 28.82 runsPP 400.0 (one run)llama.cpp4K to 200K context |
| Muse 30B4-bit | 30.0one runllama.cpp8K context | 18.0one runllama.cpp132K context |
| Qwen3.6 27B4-bit | 42.0one runPP 1020.0 (one run)llama.cpp100K context | 22.0one runPP 760.0 (one run)llama.cpp75K context |
Estimated by the calculator
The quant the calculator recommends for each card and its estimated speed at a 32,768-token context with 32 GB of system RAM, from memory bandwidth and the model's shape. Where a model was also measured above, the measurement wins.
| Model | NVIDIA RTX 3090 | NVIDIA RTX 5060 Ti 16GB |
|---|---|---|
| Qwen3.8 27B | 33.9 t/sQ4_K_M | 29.2 t/sQ2_K |
| Muse 30B | 30.5 t/sQ4_K_M | 21.9 t/sQ3_K_S |
| Qwen3.6 27B | 33.9 t/sQ4_K_M | 29.2 t/sQ2_K |
| Qwen3.8 125B · 6B active | 19.4 t/sQ2_K, experts in RAM | doesn't fit at these settings |
| Qwen3.6 35B · 3B active | 120.1 t/sQ4_K_S | 96.6 t/sQ2_K |
| DeepSeek V4 Flash 284B · 13B active | doesn't fit at these settings | doesn't fit at these settings |
| Gemma 4 26B · 4B active | 76.4 t/sQ5_K_S | 65.1 t/sQ3_K_S |
| DeepSeek V4.1 Flash 552B · 16B active | doesn't fit at these settings | doesn't fit at these settings |
| GLM-5.3 320B · 18B active | doesn't fit at these settings | doesn't fit at these settings |
Change the context, RAM or model in the calculator: NVIDIA RTX 3090 · NVIDIA RTX 5060 Ti 16GB
Every report on each card, with its source: NVIDIA RTX 3090 · NVIDIA RTX 5060 Ti 16GB