will it local

Best local LLMs for the Intel UHD 620 (iGPU)

Shared-memory device — speed scales with RAM bandwidth (38.4 GB/s shared RAM), usable for inference ≈ 5.5 GB of 8 GB system RAM

Speed

computed band 1.1-3.4 tok/s for 8B Q4_K_M (roofline, 38.4 GB/s shared RAM)

Effective decode window: 0.15–0.45 of 38.4 GB/s nominal → ~10–17 GB/s effective (llama.cpp decode, Q4_K_M basis; card-specific window from cited community benches; nominal is config-arithmetic shared-RAM bandwidth)

What fits (computed)

→ 13b-q4 no 14.6 GB vs 5.5 GB usable — does not fit

→ 3b-q4 full 2.8 GB needed of 5.5 GB usable — headroom for context

→ 4b-q4 full 3.7 GB needed of 5.5 GB usable — headroom for context

→ 8b-q4 offload-partial 6.1 GB vs 5.5 GB usable — partial CPU offload, expect large speed loss

Cited community benches

no single-card bench published — band is computed (see notes)

Runs fully in memory (machine alternatives)

Nominal bandwidth: 38.4 GB/s shared memory — 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)

→ qwen3:4b on Raspberry Pi 5 + AI Kit (Hailo-8L) ($110) 4B · 2.6GB · needs ~3.2GB (4k ctx)

→ llama3.2:1b on Raspberry Pi 5 + AI Kit (Hailo-8L) ($110) 1B · 1.3GB · needs ~1.4GB (4k ctx)

→ llama3.2:3b on Raspberry Pi 5 + AI Kit (Hailo-8L) ($110) 3B · 2.0GB · needs ~2.4GB (4k ctx)

→ gemma3:1b on Raspberry Pi 5 + AI Kit (Hailo-8L) ($110) 1B · 1.0GB · needs ~1.1GB (4k ctx)

→ phi4:mini on Raspberry Pi 5 + AI Kit (Hailo-8L) ($110) mini 3.8B · 2.5GB · needs ~3GB (4k ctx)

→ moondream:2b on Raspberry Pi 5 + AI Kit (Hailo-8L) ($110) 2B · 1.7GB · needs ~3.7GB (4k ctx)

→ whisper:tiny on Raspberry Pi 5 + AI Kit (Hailo-8L) ($110) tiny · 0.1GB · needs ~2.1GB (4k ctx)

→ whisper on Raspberry Pi 5 + AI Kit (Hailo-8L) ($110) base · 0.3GB · needs ~2.3GB (4k ctx)

→ whisper:small on Raspberry Pi 5 + AI Kit (Hailo-8L) ($110) small · 0.9GB · needs ~2.9GB (4k ctx)

→ whisper:medium on Raspberry Pi 5 + AI Kit (Hailo-8L) ($110) medium · 3.1GB · needs ~5.1GB (4k ctx)

→ kokoro:82m on Raspberry Pi 5 + AI Kit (Hailo-8L) ($110) 82M · 0.3GB · needs ~2.3GB (4k ctx)

→ gemma3:4b on Raspberry Pi 5 + AI Kit (Hailo-8L) ($110) 4B · 3.3GB · needs ~3.8GB (4k ctx)

→ deepseek-r1:7b on Raspberry Pi 5 + AI Kit (Hailo-8L) ($110) distill 7B · 4.7GB · needs ~4.9GB (4k ctx)

→ mistral:7b on Raspberry Pi 5 + AI Kit (Hailo-8L) ($110) 7B · 4.1GB · needs ~4.6GB (4k ctx)

→ llama3.1:8b on Tesla P100 (16GB, used) ($135) 8B · 4.9GB · needs ~5.4GB (4k ctx)

→ deepseek-r1:8b on Tesla P100 (16GB, used) ($135) distill 8B · 4.9GB · needs ~5.4GB (4k ctx)

Borderline — runs, but offloads

→ qwen3:8b 8B — 5.8 of 5.5 GB usable — barely over; real with q8 KV cache (halves KV) or shorter context

→ qwen2.5vl:7b 7B — 5.8 of 5.5 GB usable — barely over; real with q8 KV cache (halves KV) or shorter context

→ llava:7b 7B — 6.7 GB vs 5.5 GB usable — partial CPU offload, expect large speed loss

Gotchas

⚠ 24 EUs on shared DDR4 — decode is effectively CPU speed; SYCL/Vulkan backends barely move it

⚠ Kaby Lake-R through Comet Lake laptops; usually paired with 8-16GB

⚠ 0.6-1B Q4 toys only

Where to go next

—