will it local
2GB VRAM wall: nothing beyond 1-2B models fits on-GPU Maxwell is dropped from current CUDA toolkits — pin older llama.cpp builds or run Vulkan backend Mobile GM206: check for throttle on sustained load
⚠ Laptop part — mobile clocks and thermals run below desktop sibling speeds; treat the band as optimistic for sustained load.
→explore the full catalogupgrade paths with live prices · what fits on each machinecomputed band 4.4-7.8 tok/s for 8B Q4_K_M (roofline, 80.19 GB/s VRAM)
Effective decode window: 0.28–0.50 of 80.19 GB/s nominal → ~22–40 GB/s effective (llama.cpp decode, Q4_K_M basis; arch window, community-calibrated)
→ 13b-q4 no 14.6 GB vs 2 GB usable — does not fit
→ 3b-q4 no 2.8 GB vs 2 GB usable — does not fit
→ 4b-q4 no 3.7 GB vs 2 GB usable — does not fit
→ 8b-q4 no 6.1 GB vs 2 GB usable — does not fit
no single-card bench published — band is computed (see notes)
Nominal bandwidth: 80.19 GB/s — see the effective decode window above
→ qwen3:0.6b on Raspberry Pi 5 + AI Kit (Hailo-8L) ($110) 0.6B · 0.6GB · needs ~1GB (4k ctx)
→ qwen3:1.7b on Raspberry Pi 5 + AI Kit (Hailo-8L) ($110) 1.7B · 1.4GB · needs ~1.8GB (4k ctx)
→ llama3.2:1b on Raspberry Pi 5 + AI Kit (Hailo-8L) ($110) 1B · 1.3GB · needs ~1.4GB (4k ctx)
→ gemma3:1b on Raspberry Pi 5 + AI Kit (Hailo-8L) ($110) 1B · 1.0GB · needs ~1.1GB (4k ctx)
→ llama3.2:3b 3B — 2.4 GB vs 2 GB usable — partial CPU offload, expect large speed loss
→ whisper:tiny tiny — 2.1 of 2 GB usable — barely over; real with q8 KV cache (halves KV) or shorter context
→ whisper base — 2.3 GB vs 2 GB usable — partial CPU offload, expect large speed loss
→ kokoro:82m 82M — 2.3 GB vs 2 GB usable — partial CPU offload, expect large speed loss
⚠ 2GB VRAM wall: nothing beyond 1-2B models fits on-GPU
⚠ Maxwell is dropped from current CUDA toolkits — pin older llama.cpp builds or run Vulkan backend
⚠ Mobile GM206: check for throttle on sustained load
—