NVIDIA · 12GB · 3 reports
Use the calculator to check which models fit in 12 GB of VRAM. The reports below include different GPU counts and memory setups; a model listed here may need extra cards or CPU offloading.
How to compare these reports →RTX 5070 Ti Laptop 12GB · Unsloth Studio · 8,192 ctx
Reported by the source; GPU count, offloading and concurrent requests can change this figure. Check the full setup before comparing.
MTP acceptance 78-83%. Longer response 3.26 t/s, short factual 4.42 t/s, coding 4.53 t/s. Prompt processing 20-27 t/s. Model size ~17.9GB, offloading to CPU.
RTX 5070 Ti Laptop 12GB · llama.cpp · 4,096 ctx
Reported by the source; GPU count, offloading and concurrent requests can change this figure. Check the full setup before comparing.
Compared Qwen3.8 27B dense at Q2 and Q3 against Qwen3.6 35B-A3B MoE on 12GB VRAM. MoE was fastest and passed sanity test; dense Q3 was slow (7.5-9.1 t/s). Author prefers MoE for local use.
RTX 5070 Ti Laptop 12GB · llama.cpp · 262,144 ctx
Reported by the source; GPU count, offloading and concurrent requests can change this figure. Check the full setup before comparing.
User runs Qwen 3.8 27B UD Q4_K_XL on RTX 5070 Ti Mobile 12GB with llama.cpp. Two configs: context-prioritized (ctx 262144, ~1.5-5 t/s) and speed-prioritized (ctx 98304, ~9-11.5 t/s). KV cache q8_0. Praises model's capability and instruction following.