local intelligence

will it local

Best local LLMs for the NVIDIA GeForce RTX 2080 Ti

11GB · Turing TU102

The used-market legend: 11GB + 616 GB/s at used-market prices — the fastest pre-3000 Turing card and a real 13B/14B machine GDDR6X-era thermals do NOT apply (this is GDDR6), but early TU102 cards had memory failures — check VRAM temps and warranty void stickers on used units Build llama.cpp with CMAKE_CUDA_ARCHITECTURES=75 or silently lose ~19% to PTX JIT Dual 2080 Ti tensor-parallel runs 27B at ~59.6 tok/s measured — the multi-GPU stack play starts here

explore the full catalogupgrade paths with live prices · what fits on each machine

Speed

computed band 58.0-86.0 tok/s for 8B Q4_K_M (roofline, 616.0 GB/s VRAM); community: 72.9 tok/s Llama 3.1 8B Instruct Q4_K_M (https://www.localscore.ai/accelerator/112); community: 39.5 tok/s Qwen2.5 14B Instruct Q4_K_M (https://www.localscore.ai/accelerator/112); community: 59.6 tok/s Qwen3.6 27B (dual modded 2080 Ti, tensor parallel) Q4 (https://ai-muninn.com/en/blog/qwen38-dual-2080ti-tensor-parallel)

Effective decode window: 0.48–0.72 of 616 GB/s nominal → ~296–444 GB/s effective (llama.cpp decode, Q4_K_M basis; arch window, community-calibrated)

What fits (computed)

13b-q4 no 14.6 GB vs 11 GB usable — does not fit

3b-q4 full 2.8 GB needed of 11 GB usable — headroom for context

4b-q4 full 3.7 GB needed of 11 GB usable — headroom for context

8b-q4 full 6.1 GB needed of 11 GB usable — headroom for context

Cited community benches (3 rows)

→ 72.9 tok/s · Llama 3.1 8B Instruct Q4_K_M LocalScore accelerator/112 · source

→ 39.5 tok/s · Qwen2.5 14B Instruct Q4_K_M LocalScore accelerator/112 · source

→ 59.6 tok/s · Qwen3.6 27B (dual modded 2080 Ti, tensor parallel) Q4 ai-muninn measured (2x GPU) · source

Runs fully in memory (machine alternatives)

Nominal bandwidth: 616.0 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)

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)

llava:7b on Raspberry Pi 5 + AI Kit (Hailo-8L) ($110) 7B · 4.7GB · needs ~6.7GB (4k ctx)

qwen3:8b on Tesla P100 (16GB, used) ($135) 8B · 5.2GB · needs ~5.8GB (4k ctx)

qwen2.5vl:7b on Tesla P100 (16GB, used) ($135) 7B · 5.6GB · needs ~5.8GB (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)

whisper:large-v3 on Tesla P100 (16GB, used) ($135) large-v3 · 6.2GB · needs ~8.2GB (4k ctx)

mistral-nemo:12b on Tesla P100 (16GB, used) ($135) 12B · 7.1GB · needs ~7.7GB (4k ctx)

sdxl on Tesla P100 (16GB, used) ($135) SDXL base · 6.9GB · needs ~8.9GB (4k ctx)

gemma3:12b on Tesla P100 (16GB, used) ($135) 12B · 8.1GB · needs ~9.6GB (4k ctx)

qwen3:14b on Tesla P100 (16GB, used) ($135) 14B · 9.3GB · needs ~9.9GB (4k ctx)

deepseek-r1:14b on Tesla P100 (16GB, used) ($135) distill 14B · 9.0GB · needs ~9.8GB (4k ctx)

phi4:14b on Tesla P100 (16GB, used) ($135) 14B · 9.1GB · needs ~9.9GB (4k ctx)

Borderline — runs, but offloads

→ llava:13b 13B — 11.1 of 11 GB usable — barely over; real with q8 KV cache (halves KV) or shorter context

Gotchas

⚠ The used-market legend: 11GB + 616 GB/s at used-market prices — the fastest pre-3000 Turing card and a real 13B/14B machine

⚠ GDDR6X-era thermals do NOT apply (this is GDDR6), but early TU102 cards had memory failures — check VRAM temps and warranty void stickers on used units

⚠ Build llama.cpp with CMAKE_CUDA_ARCHITECTURES=75 or silently lose ~19% to PTX JIT

⚠ Dual 2080 Ti tensor-parallel runs 27B at ~59.6 tok/s measured — the multi-GPU stack play starts here

Where to go next

rtx-3090

24GB used value king — 32B Q4 fits fully

→ its best-models page

rtx-3080-ti

912 GB/s +1GB, far less power

→ its best-models page

2026-09-18 · ← full hardware catalog · fit = working set vs VRAM · speeds are community-reported, cited in our knowledge base