DeskGPU
English

NVIDIA GeForce GTX 1660 SUPER: which LLMs can it run?

This page was written and translated automatically by AI and checked by software against its sources, with no human editing. Specifications come from the sources listed; memory requirements and speeds are estimates calculated with published formulas. How we use AI

NVIDIA GeForce GTX 1660 SUPER has 6 GB of memory (manufacturer figure). Below, every model in the database is checked against it with llama.cpp at a context length of 8,192 tokens (or the model’s maximum if shorter).

Values updated Oct 19, 2026

Specifications

Memory (manufacturer figure)
6 GB
Memory type
GDDR6
Memory bus width
192 bit
Memory bandwidth
336 GB/s
Board power (TDP)
125 W
FP16 tensor, dense
Not shown: open sources do not give two independent, agreeing values with a manufacturer reference.
BF16 tensor, dense
Not shown: open sources do not give two independent, agreeing values with a manufacturer reference.
FP8 tensor, dense
Not shown: open sources do not give two independent, agreeing values with a manufacturer reference.
FP4 tensor, dense
Not shown: open sources do not give two independent, agreeing values with a manufacturer reference.
INT8 tensor, dense
Not shown: open sources do not give two independent, agreeing values with a manufacturer reference.
Interface
Not shown: open sources do not give two independent, agreeing values with a manufacturer reference.

Each value lists every source with its own figure; units have been normalized. A manufacturer’s “GB” of GPU memory is counted as GiB (2³⁰ bytes), which is also how AMD’s ROCm table states it. How specifications are checked

Run profile

Run profile: llama.cpp, CUDA back end (NVIDIA discrete GPU), all layers on the GPU, Flash Attention on, micro-batch (-ub) 512, F16 KV cache, context (-c) as shown; “Fits” is judged with llama-server’s default 4 parallel sequences (-np auto, unified KV cache) and “Doesn’t fit” with 1 sequence; idle GPU with no other programs using its memory.

Memory counted as available for “Fits” (C_low): 2.88 GiB per card = 6 GiB nominal − 1,054 MiB (the largest gap between nominal and reported total measured on consumer and workstation GPUs in this profile, used because no public log reports this GPU’s own total) − 2,112 MiB (the largest start-up and runtime use seen in this profile’s public logs), rounded down to a multiple of 64 MiB.

These results assume the card’s default configuration. Settings that reduce usable memory (for example ECC, where the card supports it) are not covered.

Memory used for “Doesn’t fit” (C_high): the full 6 GiB per card. Run profiles and their measurements

Which models fit

“Fits” means llama.cpp's default --fit load check would place all layers on the GPU: our estimate of llama.cpp's buffers, plus its error margin, plus llama.cpp's 1 GiB per-card reserve, is within the memory counted as available. It does not promise that peak memory while generating fits. How results are decided

Context 8,192 tokens (or the model’s maximum if shorter). The last columns show the longest context at that quantization and the bandwidth-based speed ceiling.
Model Parameters Highest quantization that fits Available memory needed for “Fits” Longest context Speed ceiling Result
gemma-4-E2B 4.6B Q3_K_M 2.9 GiB 9,728 ≤ 255 tok/s Fits
gemma-4-E2B-it 4.6B Q3_K_M 2.8 GiB 16,896 ≤ 267 tok/s Fits
Apertus-v1.1-4B 3.8B IQ2_XXS 2.7 GiB 4,096 ≤ 228 tok/s Fits
BitCPM-CANN-3B-unquantized 3.6B Q2_K 2.9 GiB 8,704 ≤ 209 tok/s Fits
Qwen2.5-3B-Instruct 3.4B Q2_K 2.9 GiB 8,448 ≤ 214 tok/s Fits
AI21-Jamba-Reasoning-3B 3.2B Q3_K_M 2.8 GiB 45,312 ≤ 228 tok/s Fits
AI21-Jamba2-3B 3.2B Q3_K_M 2.8 GiB 45,312 ≤ 228 tok/s Fits
SmolLM3-3B-GSM8K-SFT 3.1B IQ2_XXS 2.8 GiB 9,472 ≤ 220 tok/s Fits
LFM2-2.6B-Exp 2.7B Q4_0 2.9 GiB 8,704 ≤ 208 tok/s Fits
LFM2.5-2.6B-Base 2.7B Q3_K_M 2.7 GiB 15,360 ≤ 225 tok/s Fits
EuroMoE-2.6B-A0.6B-Instruct-2512 2.6B (498M active) Q3_K_M 2.7 GiB 13,056 ≤ 685 tok/s Fits
LFM2-2.6B-Longevity 2.6B IQ4_XS 2.8 GiB 12,288 ≤ 217 tok/s Fits
LFM2-2.6B-Transcript 2.6B Q4_0 2.9 GiB 8,704 ≤ 208 tok/s Fits
LFM2-VL-3B 2.6B Q4_0 2.9 GiB 8,704 ≤ 208 tok/s Fits
MiniCPM5-2B 2.5B Q3_K_M 2.7 GiB 12,288 ≤ 234 tok/s Fits
MiniCPM5-2B-Base 2.5B Q3_K_M 2.8 GiB 10,240 ≤ 221 tok/s Fits
HiPO-1.7B 2.0B IQ2_XXS 2.8 GiB 8,704 ≤ 220 tok/s Fits
Youtu-LLM-2B 2.0B Q4_K_M 2.8 GiB 10,496 ≤ 220 tok/s Fits
Youtu-LLM-2B-Base 2.0B Q4_K_M 2.8 GiB 10,496 ≤ 220 tok/s Fits
Qwen3.5-2B 1.9B Q4_0 2.9 GiB 8,192 ≤ 258 tok/s Fits
Qwen3.5-2B-Base 1.9B Q4_0 2.9 GiB 8,960 ≤ 260 tok/s Fits
granite-4.0-1b 1.6B Q3_K_M 2.8 GiB 9,216 ≤ 220 tok/s Fits
Apertus-v1.1-1.5B 1.5B Q8_0 2.7 GiB 4,096 ≤ 231 tok/s Fits
Apertus-v1.1-1.5B-Instruct 1.5B Q8_0 2.7 GiB 4,096 ≤ 231 tok/s Fits
granite-4.0-h-1b 1.5B Q6_K 2.8 GiB 20,480 ≤ 265 tok/s Fits
Llama-3.2-1B-Instruct 1.2B Q8_0 2.8 GiB 9,472 ≤ 212 tok/s Fits
LFM2.5-1.2B-Base 1.2B Q8_0 2.6 GiB 27,648 ≤ 250 tok/s Fits
LFM2.5-1.2B-Instruct 1.2B Q8_0 2.6 GiB 27,648 ≤ 250 tok/s Fits
LFM2.5-1.2B-JP 1.2B Q8_0 2.6 GiB 27,648 ≤ 250 tok/s Fits
LFM2.5-1.2B-JP-202606 1.2B Q8_0 2.6 GiB 27,648 ≤ 250 tok/s Fits
LFM2.5-1.2B-Thinking 1.2B Q8_0 2.6 GiB 27,648 ≤ 250 tok/s Fits
LFM2.5-VL-1.6B 1.2B Q8_0 2.6 GiB 27,648 ≤ 250 tok/s Fits
LFM2.5-VL-1.6B-Extract 1.2B Q8_0 2.6 GiB 27,648 ≤ 250 tok/s Fits
MiniCPM5-1B 1.1B Q6_K 2.2 GiB 34,560 ≤ 364 tok/s Fits
MiniCPM5-1B-Base 1.1B Q6_K 2.2 GiB 34,560 ≤ 364 tok/s Fits
MiniCPM5-1B-SFT 1.1B Q6_K 2.2 GiB 34,560 ≤ 364 tok/s Fits
Holo-3.1-0.8B 1.0B Q6_K 2.3 GiB 48,896 ≤ 467 tok/s Fits
MiniCPM-V-4.7-1B 1.0B Q8_0 2.5 GiB 36,352 ≤ 373 tok/s Fits
gemma-3-1b-it 1.0B Q8_0 2.7 GiB 32,768 ≤ 303 tok/s Fits
Qwen3.5-0.8B 752M Q8_0 2.5 GiB 36,352 ≤ 373 tok/s Fits
Qwen3.5-0.8B-Base 752M Q6_K 2.3 GiB 48,896 ≤ 467 tok/s Fits
MiniCPM-V-4.6 752M Q8_0 2.5 GiB 36,352 ≤ 373 tok/s Fits
MiniCPM-V-4.6-Thinking 752M Q8_0 2.5 GiB 36,352 ≤ 373 tok/s Fits
Qwen3-0.6B 752M Q8_0 2.9 GiB 8,192 ≤ 214 tok/s Fits
HunyuanOCR 539M Q6_K 2.5 GiB 11,776 ≤ 269 tok/s Fits
LFM2.5-350M 354M BF16 2.1 GiB 65,024 ≤ 415 tok/s Fits
LFM2.5-Encoder-350M-Diffusion 354M Q6_K 1.7 GiB 94,208 ≤ 858 tok/s Fits
LFM2.5-VL-450M 354M BF16 2.1 GiB 65,024 ≤ 415 tok/s Fits
LFM2.5-VL-450M-Extract 354M F16 2.1 GiB 65,024 ≤ 415 tok/s Fits
granite-4.0-350m 352M BF16 2.2 GiB 30,464 ≤ 358 tok/s Fits
granite-4.0-h-350m 340M BF16 2.1 GiB 32,768 ≤ 470 tok/s Fits
functiongemma-270m-it 268M BF16 2.1 GiB 32,768 ≤ 590 tok/s Fits
LFM2.5-230M 230M BF16 1.8 GiB 82,432 ≤ 600 tok/s Fits

Borderline models

Within the estimate error range: it may fit or it may not. Try a shorter context or a lower quantization.

Context 8,192 tokens. No quantization of these models fits; the one shown is the highest that is borderline.
Model Parameters Highest borderline quantization Available memory needed for “Fits” Result
FrogMini-14B-2510 14.8B IQ2_M 7.5 GiB Tight
Nemotron-Cascade-14B-Thinking 14.8B IQ2_M 7.5 GiB Tight
Nemotron-Terminal-14B 14.8B IQ2_M 7.5 GiB Tight
NousCoder-14B 14.8B IQ2_M 7.5 GiB Tight
Qwen3-Nemotron-14B-BRRM 14.8B IQ2_M 7.5 GiB Tight
WebWorld-14B 14.8B IQ2_M 7.5 GiB Tight
Apriel-1.6-15b-Thinker 14.4B IQ2_M 7.6 GiB Tight
Mistral-Nemo-Instruct-2407 12.2B Q2_K 7.1 GiB Tight
Mellum2-12B-A2.5B-Base 12.1B (2.2B active) IQ4_XS 8.0 GiB Tight
Mellum2-12B-A2.5B-Instruct 12.1B (2.2B active) Q3_K_M 7.6 GiB Tight
Mellum2-12B-A2.5B-Instruct-SFT 12.1B (2.2B active) IQ4_XS 8.0 GiB Tight
Mellum2-12B-A2.5B-Thinking 12.1B (2.2B active) Q3_K_M 7.6 GiB Tight
Mellum2-12B-A2.5B-Thinking-SFT 12.1B (2.2B active) IQ4_XS 8.0 GiB Tight
gemma-4-12B-it 11.9B Q3_K_M 8.7 GiB Tight
gemma-3-12b-it 11.8B Q2_K 8.2 GiB Tight
translategemma-12b-it 11.8B Q2_K 8.2 GiB Tight
AutoGLM-Phone-9B 9.4B Q4_K_M 7.3 GiB Tight
AutoGLM-Phone-9B-Multilingual 9.4B Q4_K_M 7.3 GiB Tight
GLM-4.6V-Flash 9.4B Q4_K_M 7.3 GiB Tight
gemma-2-9b-it 9.2B IQ3_XXS 8.0 GiB Tight
EuroLLM-9B-Instruct-2512 9.2B Q4_K_M 7.8 GiB Tight
Fara1.5-9B 9.0B Q5_K_M 8.0 GiB Tight
Holo-3.1-9B 9.0B Q5_K_M 7.6 GiB Tight
MiMo-V2.6-Distill-Qwen-9B 9.0B Q5_K_M 8.0 GiB Tight
Qwen3.5-9B 9.0B Q5_K_M 7.7 GiB Tight
Qwen3.5-9B-Base 9.0B Q5_K_M 7.6 GiB Tight
UI-Mate-9B 9.0B Q5_K_M 8.0 GiB Tight
UI-Venus-2-9B 9.0B Q5_K_M 8.0 GiB Tight
NVIDIA-Nemotron-Nano-9B-v2 8.9B Q4_K_M 8.0 GiB Tight
granite-4.1-8b 8.8B Q4_K_M 7.6 GiB Tight
granite-4.2-8b 8.8B Q4_K_M 7.6 GiB Tight
Ministral-3-8B-Instruct-2512 8.5B Q4_K_M 7.2 GiB Tight
Ministral-3-8B-Reasoning-2512 8.5B Q4_K_M 7.2 GiB Tight
LFM2.5-8B-A1B 8.5B (1.7B active) Q5_K_M 7.3 GiB Tight
granite-guardian-4.1-8b 8.4B Q4_K_M 7.6 GiB Tight
LFM2-8B-A1B 8.3B (1.6B active) Q5_K_M 7.2 GiB Tight
rnj-1-instruct 8.3B Q4_K_M 7.3 GiB Tight
AesCode-8B 8.2B Q4_K_M 7.3 GiB Tight
ArmorOCR 8.2B Q4_K_M 7.1 GiB Tight
AstaBrief_8B 8.2B Q5_K_M 7.8 GiB Tight
ContextPilot-8B 8.2B Q5_K_M 7.8 GiB Tight
Cosmos-Reason2-8B 8.2B Q5_K_M 7.8 GiB Tight
DeepSeek-R1-0528-Qwen3-8B 8.2B Q5_K_M 7.8 GiB Tight
Holo2-8B 8.2B Q5_K_M 7.8 GiB Tight
intent-aware-lfqa-qwen3-8b-baseline 8.2B Q5_K_M 7.8 GiB Tight
intent-aware-lfqa-qwen3-8b-intent-explicit 8.2B Q5_K_M 7.8 GiB Tight
intent-aware-lfqa-qwen3-8b-intent-implicit 8.2B Q5_K_M 7.8 GiB Tight
intent-aware-lfqa-qwen3-8b-multiview 8.2B Q5_K_M 7.8 GiB Tight
JanusCoder-8B 8.2B Q5_K_M 7.8 GiB Tight
MathForm-8B 8.2B Q5_K_M 7.8 GiB Tight
Nemotron-Cascade-8B 8.2B Q5_K_M 7.8 GiB Tight
Nemotron-Cascade-8B-Thinking 8.2B Q5_K_M 7.8 GiB Tight
Nemotron-Orchestrator-8B 8.2B Q5_K_M 7.8 GiB Tight
Nemotron-Terminal-8B 8.2B Q5_K_M 7.8 GiB Tight
Qwen3-8B 8.2B Q5_K_M 7.8 GiB Tight
Qwen3-Nemotron-8B-BRRM 8.2B Q5_K_M 7.8 GiB Tight
Qwen3-VL-8B-Instruct 8.2B Q5_K_M 7.8 GiB Tight
Qwen3-VL-8B-Thinking 8.2B Q5_K_M 7.8 GiB Tight
RLVR-8B-0926 8.2B Q5_K_M 7.8 GiB Tight
Step3-VL-10B-Base 8.2B Q5_K_M 7.8 GiB Tight
UI-Venus-1.5-8B 8.2B Q5_K_M 7.8 GiB Tight
WebWorld-8B 8.2B Q5_K_M 7.8 GiB Tight
ZwZ-8B 8.2B Q5_K_M 7.8 GiB Tight
MiniCPM-o-4_5 8.2B Q5_K_M 7.8 GiB Tight
granite-3.3-8b-instruct 8.2B Q4_K_M 7.4 GiB Tight
granite-guardian-3.2-8b-factuality-detection 8.2B Q4_K_M 7.4 GiB Tight
intent-aware-lfqa-llama3-8b-baseline 8.0B Q5_K_M 7.6 GiB Tight
intent-aware-lfqa-llama3-8b-intent-explicit 8.0B Q5_K_M 7.6 GiB Tight
intent-aware-lfqa-llama3-8b-intent-implicit 8.0B Q5_K_M 7.6 GiB Tight
intent-aware-lfqa-llama3-8b-multiview 8.0B Q5_K_M 7.6 GiB Tight
DeepSeek-R1-Distill-Llama-8B 8.0B Q5_K_M 7.6 GiB Tight
HARC-Llama-3.1-8B-Instruct 8.0B Q5_K_M 7.6 GiB Tight
Llama-3.1-8B-Instruct 8.0B Q5_K_M 7.6 GiB Tight
AprielGuard 7.9B Q6_K 7.9 GiB Tight
MiMo-Embodied-7B 7.6B Q5_K_M 7.4 GiB Tight
DeepSeek-R1-Distill-Qwen-7B 7.6B Q6_K 7.5 GiB Tight
Fara-7B 7.6B Q6_K 7.5 GiB Tight
HARC-Qwen2.5-7B-Instruct 7.6B Q6_K 7.5 GiB Tight
JanusCoderV-7B 7.6B Q6_K 7.5 GiB Tight
olmOCR-2-7B-1025 7.6B Q6_K 7.5 GiB Tight
Qwen2.5-7B-Instruct 7.6B Q6_K 7.5 GiB Tight
Qwen2.5-Coder-7B-Instruct 7.6B Q6_K 7.5 GiB Tight
Qwen2.5-VL-7B-Instruct 7.6B Q6_K 7.5 GiB Tight
Qwen2.5-VL-7B-Surg-CholecT50 7.6B Q6_K 7.5 GiB Tight
Spatial-SSRL-7B 7.6B Q6_K 7.5 GiB Tight
X-Reasoner-7B 7.6B Q6_K 7.5 GiB Tight
ZwZ-7B 7.6B Q6_K 7.5 GiB Tight
gemma-4-E4B 7.5B Q8_0 6.7 GiB Tight
gemma-4-E4B-it 7.5B Q8_0 6.9 GiB Tight
Bielik-Minitron-7B-v3.0-Instruct 7.5B Q5_K_M 7.7 GiB Tight
ContextPilot-E4B 7.5B Q8_0 6.6 GiB Tight
Olmo-3.1-7B-RL-Zero-Code 7.3B IQ4_XS 9.2 GiB Tight
Olmo-3.1-7B-RL-Zero-Math 7.3B IQ4_XS 9.2 GiB Tight
Olmo-3-7B-Instruct 7.3B IQ4_XS 9.2 GiB Tight
Olmo-3-7B-Instruct-DPO 7.3B IQ4_XS 9.2 GiB Tight
Olmo-3-7B-Instruct-SFT 7.3B IQ4_XS 9.2 GiB Tight
Olmo-3-7B-RL-Zero-Code 7.3B IQ4_XS 9.2 GiB Tight
Olmo-3-7B-RL-Zero-General 7.3B IQ4_XS 9.2 GiB Tight
Olmo-3-7B-RL-Zero-IF 7.3B IQ4_XS 9.2 GiB Tight
Olmo-3-7B-RL-Zero-Math 7.3B IQ4_XS 9.2 GiB Tight
Olmo-3-7B-RL-Zero-Mix 7.3B IQ4_XS 9.2 GiB Tight
Olmo-3-7B-Think 7.3B IQ4_XS 9.2 GiB Tight
Olmo-3-7B-Think-DPO 7.3B IQ4_XS 9.2 GiB Tight
Olmo-3-7B-Think-SFT 7.3B IQ4_XS 9.2 GiB Tight
Mistral-7B-Instruct-v0.2 7.2B Q5_K_M 7.3 GiB Tight
Trinity-Nano-Base 6.1B (819M active) Q6_K 6.7 GiB Tight
Trinity-Nano-Base-Pre-Anneal 6.1B (819M active) Q6_K 6.4 GiB Tight
Trinity-Nano-Preview 6.1B (819M active) Q8_0 8.1 GiB Tight
Youtu-VL-4B-Instruct 4.9B Q6_K 5.9 GiB Tight
Holo-3.1-4B 4.8B Q8_0 6.4 GiB Tight
AgentCPM-Explore 4.4B Q6_K 5.7 GiB Tight
CapRL-Qwen3VL-4B 4.4B Q8_0 6.7 GiB Tight
Holo2-4B 4.4B Q8_0 6.7 GiB Tight
intent-aware-lfqa-qwen3-4b-baseline 4.4B Q8_0 6.7 GiB Tight
intent-aware-lfqa-qwen3-4b-intent-explicit 4.4B Q8_0 6.7 GiB Tight
intent-aware-lfqa-qwen3-4b-intent-implicit 4.4B Q8_0 6.7 GiB Tight
intent-aware-lfqa-qwen3-4b-multiview 4.4B Q8_0 6.7 GiB Tight
Spatial-SSRL-Qwen3VL-4B 4.4B Q8_0 6.7 GiB Tight
ZwZ-4B 4.4B Q6_K 5.7 GiB Tight
Fara1.5-4B 4.2B Q8_0 6.4 GiB Tight
FrogNano-4B-2609 4.2B Q8_0 6.4 GiB Tight
Qwen3.5-4B 4.2B Q8_0 6.4 GiB Tight
Qwen3.5-4B-Base 4.2B Q8_0 6.4 GiB Tight
UserRepGen-4B 4.2B Q8_0 6.4 GiB Tight
Riva-Translate-4B-Instruct-v1.1 4.2B Q6_K 6.0 GiB Tight
GELab-Zero-4B-preview 4.0B Q8_0 6.7 GiB Tight
Qwen3-4B 4.0B Q8_0 6.7 GiB Tight
Qwen3-4B-Instruct-2507 4.0B Q8_0 6.7 GiB Tight
Qwen3-VL-4B-Instruct 4.0B Q8_0 6.7 GiB Tight
Qwen3-VL-4B-Thinking 4.0B Q8_0 6.7 GiB Tight
NVIDIA-Nemotron-3-Nano-4B-BF16 4.0B Q8_0 5.5 GiB Tight
gemma-3-4b-it 3.9B Q8_0 6.3 GiB Tight
medgemma-1.5-4b-it 3.9B Q8_0 6.3 GiB Tight
Nemotron-3-Content-Safety 3.9B Q6_K 5.4 GiB Tight
translategemma-4b-it 3.9B Q8_0 6.3 GiB Tight
Phi-4-mini-instruct 3.8B Q8_0 6.5 GiB Tight
Phi-3.5-mini-instruct 3.8B Q6_K 7.4 GiB Tight
granite-4.2-3b 3.7B Q8_0 5.5 GiB Tight
kanana-2-3b-instruct 3.5B Q8_0 5.6 GiB Tight
Ministral-3-3B-Instruct-2512 3.4B Q8_0 5.7 GiB Tight
Ministral-3-3B-Reasoning-2512 3.4B Q8_0 5.7 GiB Tight
granite-vision-4.1-4b 3.4B Q8_0 5.5 GiB Tight
granite-4.1-3b 3.4B Q8_0 5.5 GiB Tight
granite-guardian-4.0-3b-toxicity-ja 3.4B Q6_K 4.7 GiB Tight
tiny-aya-earth 3.3B Q8_0 5.6 GiB Tight
tiny-aya-fire 3.3B Q8_0 5.6 GiB Tight
tiny-aya-global 3.3B Q8_0 5.6 GiB Tight
tiny-aya-water 3.3B Q8_0 5.6 GiB Tight
Llama-3.2-3B-Instruct 3.2B Q8_0 5.5 GiB Tight
kanana-2-3b-base 3.2B Q6_K 4.9 GiB Tight
SmolLM3-3B 3.1B BF16 7.9 GiB Tight
d1-3B 2.7B BF16 6.7 GiB Tight
LFM2.5-2.6B 2.7B BF16 6.7 GiB Tight
LFM2.5-VL-3B 2.7B BF16 6.7 GiB Tight
CapRL-Qwen3VL-2B 2.0B F16 5.6 GiB Tight
UI-Venus-1.5-2B 2.0B F16 5.6 GiB Tight
ZwZ-2B 2.0B F16 5.6 GiB Tight
Qwen3-1.7B 1.7B BF16 5.6 GiB Tight
Qwen3-VL-2B-Instruct 1.7B BF16 5.6 GiB Tight
Qwen3-VL-2B-Thinking 1.7B BF16 5.6 GiB Tight
SingGuard-2b 1.7B F16 5.6 GiB Tight

Models that don’t fit on one NVIDIA GeForce GTX 1660 SUPER

Context 8,192 tokens. Even the lowest quantization does not fit on one card; the requirement shown is for that quantization.
Model Parameters Smallest quantization Available memory needed for “Fits” Several identical cards Result
JanusCoder-14B 14.8B Q2_K 8.0 GiB 6 identical cards (default even split) are expected to fit. Fewer cards are borderline. Doesn't fit
ContextPilot-14B 14.8B IQ2_M 8.1 GiB 6 identical cards (default even split) are expected to fit. Fewer cards are borderline. Doesn't fit
Qwen3-Coder-REAP-25B-A3B 24.9B (3.0B active) IQ2_XXS 8.1 GiB 5 identical cards (default even split) are expected to fit. Fewer cards are borderline. Doesn't fit
Trinity-Mini 26.1B (3.1B active) IQ2_XXS 8.2 GiB 6 identical cards (default even split) are expected to fit. Fewer cards are borderline. Doesn't fit
phi-4 14.7B Q2_K 8.3 GiB 6 identical cards (default even split) are expected to fit. Fewer cards are borderline. Doesn't fit
DeepSeek-R1-Distill-Qwen-14B 14.8B Q2_K 8.3 GiB 7 identical cards (default even split) are expected to fit. Fewer cards are borderline. Doesn't fit
Qwen2.5-14B-Instruct 14.8B Q2_K 8.3 GiB 7 identical cards (default even split) are expected to fit. Fewer cards are borderline. Doesn't fit
Trinity-Mini-Base-Pre-Anneal 26.1B (3.1B active) IQ2_XXS 8.4 GiB 6 identical cards (default even split) are expected to fit. Fewer cards are borderline. Doesn't fit
Trinity-Mini-Base 26.1B (3.1B active) IQ2_XXS 8.7 GiB 6 identical cards (default even split) are expected to fit. Fewer cards are borderline. Doesn't fit
EuroLLM-22B-2512 22.6B IQ2_XXS 9.1 GiB 7 identical cards (default even split) are expected to fit. Fewer cards are borderline. Doesn't fit
Mellum2.1-12B-A2.5B-Thinking 12.1B (2.2B active) Q4_K_M 9.4 GiB 6 identical cards (default even split) are expected to fit. Fewer cards are borderline. Doesn't fit
Bielik-11B-v3.0-Instruct 11.2B Q4_K_M 9.4 GiB 6 identical cards (default even split) are expected to fit. Fewer cards are borderline. Doesn't fit
Muse-Glimmer-30B 27.9B IQ2_XXS 10.2 GiB Borderline with 8 cards. Doesn't fit
GLM-4.7-Flash-REAP-23B-A3B 23.0B (3.6B active) Q2_K 10.2 GiB 7 identical cards (default even split) are expected to fit. Fewer cards are borderline. Doesn't fit
Ministral-3-14B-Instruct-2512 13.5B Q4_K_M 10.3 GiB 7 identical cards (default even split) are expected to fit. Fewer cards are borderline. Doesn't fit
Ministral-3-14B-Reasoning-2512 13.5B Q4_K_M 10.3 GiB 7 identical cards (default even split) are expected to fit. Fewer cards are borderline. Doesn't fit
Holo4-27B 26.9B IQ2_XXS 10.5 GiB Borderline with 8 cards. Doesn't fit
EuroLLM-22B-Instruct-2512 22.6B IQ2_M 10.6 GiB 7 identical cards (default even split) are expected to fit. Fewer cards are borderline. Doesn't fit
Qwen3.8-27B 26.9B IQ2_XXS 10.6 GiB Borderline with 8 cards. Doesn't fit
Qwen3.8-3.6-27B-blend 26.9B IQ2_XXS 10.6 GiB Borderline with 8 cards. Doesn't fit
Kimi-Linear-REAP-35B-A3B-Instruct 35.1B (3.1B active) IQ2_XXS 10.7 GiB 8 identical cards (default even split) are expected to fit. Fewer cards are borderline. Doesn't fit
North-Mini-Code-1.0 30.5B (3.3B active) IQ2_XXS 10.7 GiB Borderline with 8 cards. Doesn't fit
Qwen3.5-35B-A3B-Base 34.7B (2.9B active) IQ2_XXS 10.9 GiB 7 identical cards (default even split) are expected to fit. Fewer cards are borderline. Doesn't fit
Qwen3-14B 14.8B Q4_K_M 11.0 GiB Borderline with 8 cards. Doesn't fit
Laguna-XS-2.1 33.4B (2.8B active) IQ2_XXS 11.1 GiB 7 identical cards (default even split) are expected to fit. Fewer cards are borderline. Doesn't fit
Ising-Calibration-1-35B-A3B 34.7B (2.9B active) IQ2_XXS 11.1 GiB 7 identical cards (default even split) are expected to fit. Fewer cards are borderline. Doesn't fit
Devstral-Small-2-24B-Instruct-2512 23.6B Q2_K 11.1 GiB 7 identical cards (default even split) are expected to fit. Fewer cards are borderline. Doesn't fit
Mistral-Small-3.2-24B-Instruct-2506 23.6B Q2_K 11.1 GiB 7 identical cards (default even split) are expected to fit. Fewer cards are borderline. Doesn't fit
UI-Mate-27B 26.9B IQ2_XXS 11.2 GiB Borderline with 8 cards. Doesn't fit
Qwen3.6-35B-A3B 34.7B (2.9B active) IQ2_XXS 11.3 GiB 8 identical cards (default even split) are expected to fit. Fewer cards are borderline. Doesn't fit
KAT-Coder-V2.5-Dev 34.7B (2.9B active) IQ2_XXS 11.3 GiB Borderline with 8 cards. Doesn't fit
Olmo-3.1-32B-Instruct-DPO 32.2B IQ2_XXS 11.8 GiB Borderline with 8 cards. Doesn't fit
Olmo-3-1125-32B 32.2B IQ2_XXS 11.8 GiB Borderline with 8 cards. Doesn't fit
Olmo-3-32B-Think-DPO 32.2B IQ2_XXS 11.8 GiB Borderline with 8 cards. Doesn't fit
Olmo-3-32B-Think-SFT 32.2B IQ2_XXS 11.8 GiB Borderline with 8 cards. Doesn't fit
Holo4-35B-A3B 34.7B (2.9B active) IQ2_XXS 12.0 GiB Borderline with 8 cards. Doesn't fit
Cosmos-Reason2-32B 32.8B IQ2_XXS 12.0 GiB Borderline with 8 cards. Doesn't fit
FrogBoss-32B-2510 32.8B IQ2_XXS 12.0 GiB Borderline with 8 cards. Doesn't fit
Nemotron-Terminal-32B 32.8B IQ2_XXS 12.0 GiB Borderline with 8 cards. Doesn't fit
SERA-32B 32.8B IQ2_XXS 12.0 GiB Borderline with 8 cards. Doesn't fit
WebWorld-32B 32.8B IQ2_XXS 12.0 GiB Borderline with 8 cards. Doesn't fit
gemma-4-26B-A4B-it 25.2B (3.8B active) IQ2_XXS 12.1 GiB 8 identical cards (default even split) are expected to fit. Fewer cards are borderline. Doesn't fit
Fara1.5-27B 26.9B IQ2_M 12.5 GiB Borderline with 8 cards. Doesn't fit
gpt-oss-20b 20.9B (3.6B active) Q2_K 12.6 GiB Borderline with 8 cards. Doesn't fit
UI-Mate-democua-27B 26.9B Q2_K 12.7 GiB Borderline with 8 cards. Doesn't fit
GLM-4.7-Flash 29.9B (3.6B active) Q2_K 12.8 GiB Borderline with 8 cards. Doesn't fit
Hermes-4.3-36B 36.2B IQ2_XXS 12.9 GiB Borderline with 8 cards. Doesn't fit
gemma-4-26B-A4B 25.2B (3.8B active) Q2_K 13.0 GiB Borderline with 8 cards. Doesn't fit
Holo2-30B-A3B 30.5B (3.0B active) Q2_K 13.0 GiB Borderline with 8 cards. Doesn't fit
Qwen3-30B-A3B-Instruct-2507 30.5B (3.0B active) Q2_K 13.0 GiB Borderline with 8 cards. Doesn't fit
Qwen3-Coder-30B-A3B-Instruct 30.5B (3.0B active) Q2_K 13.0 GiB Borderline with 8 cards. Doesn't fit
Qwen3-VL-30B-A3B-Instruct 30.5B (3.0B active) Q2_K 13.0 GiB Borderline with 8 cards. Doesn't fit
UI-Venus-1.5-30B-A3B 30.5B (3.0B active) Q2_K 13.0 GiB Borderline with 8 cards. Doesn't fit
OptiMind-SFT 20.9B (3.6B active) IQ2_M 13.1 GiB Borderline with 8 cards. Doesn't fit
Kimi-Linear-48B-A3B-Instruct 49.1B (3.1B active) IQ2_XXS 13.3 GiB Borderline with 8 cards. Doesn't fit
AesCode-32B 32.8B IQ2_XXS 13.6 GiB Borderline with 8 cards. Doesn't fit
AI21-Jamba2-Mini 51.6B (11.8B active) IQ2_XXS 13.6 GiB Borderline with 8 cards. Doesn't fit
gemma-2-27b-it 27.2B IQ2_M 13.8 GiB Borderline with 8 cards. Doesn't fit
granite-4.1-30b 28.9B Q2_K 13.9 GiB Borderline with 8 cards. Doesn't fit
granite-4.2-30b 29.3B Q2_K 13.9 GiB Borderline with 8 cards. Doesn't fit
Kimi-Linear-48B-A3B-Base 49.1B (3.1B active) IQ2_XXS 14.2 GiB Borderline with 8 cards. Doesn't fit
gemma-3-27b-it 27.0B Q2_K 14.4 GiB Borderline with 8 cards. Doesn't fit
translategemma-27b-it 27.0B Q2_K 14.4 GiB Borderline with 8 cards. Doesn't fit
Holo-3.1-35B-A3B 34.7B (2.9B active) Q2_K 14.4 GiB Borderline with 8 cards. Doesn't fit
Holo3-35B-A3B 34.7B (2.9B active) Q2_K 14.4 GiB Borderline with 8 cards. Doesn't fit
LFM2-24B-A2B 23.8B (2.3B active) Q4_0 14.6 GiB Borderline with 8 cards. Doesn't fit
LFM2.5-24B-A2B 24.0B (2.5B active) Q4_0 14.7 GiB Borderline with 8 cards. Doesn't fit
Olmo-3.1-32B-Instruct 32.2B Q2_K 15.0 GiB Borderline with 8 cards. Doesn't fit
Olmo-3.1-32B-Think 32.2B Q2_K 15.0 GiB Borderline with 8 cards. Doesn't fit
Olmo-3-32B-Think 32.2B Q2_K 15.0 GiB Borderline with 8 cards. Doesn't fit
DeepSeek-R1-Distill-Qwen-32B 32.8B Q2_K 15.2 GiB Borderline with 8 cards. Doesn't fit
Qwen2.5-32B-Instruct 32.8B Q2_K 15.2 GiB Borderline with 8 cards. Doesn't fit
Qwen2.5-Coder-32B-Instruct 32.8B Q2_K 15.2 GiB Borderline with 8 cards. Doesn't fit
Qwen3-VL-32B-Instruct 32.8B Q2_K 15.3 GiB Borderline with 8 cards. Doesn't fit
Qwen3-VL-32B-Thinking 32.8B Q2_K 15.3 GiB Borderline with 8 cards. Doesn't fit
Qwen3.5-27B 26.9B Q3_K_M 15.3 GiB Borderline with 8 cards. Doesn't fit
Qwen3.6-27B 26.9B Q3_K_M 15.4 GiB Borderline with 8 cards. Doesn't fit
gemma-4-12B 11.9B Q8_0 15.6 GiB Borderline with 8 cards. Doesn't fit
Ising-Calibration-1.5-31B-BF16 30.7B IQ2_XXS 16.5 GiB Borderline with 8 cards. Doesn't fit
Qwen3.5-35B-A3B 34.7B (2.9B active) Q3_K_M 17.6 GiB Borderline with 8 cards. Doesn't fit
gemma-4-31B 30.7B Q2_K 17.6 GiB Borderline with 8 cards. Doesn't fit
Holotron4-30B-A3B 31.6B (3.2B active) IQ2_M 18.9 GiB Borderline with 8 cards. Doesn't fit
NVIDIA-Nemotron-Labs-3-Elastic-30B-A3B-BF16 31.6B (3.2B active) IQ2_M 18.9 GiB Borderline with 8 cards. Doesn't fit
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-Base-BF16 31.6B (3.2B active) Q2_K 18.9 GiB Borderline with 8 cards. Doesn't fit
Nemotron-Cascade-2-30B-A3B 31.6B (3.2B active) IQ2_M 19.1 GiB Borderline with 8 cards. Doesn't fit
NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16 31.6B (3.2B active) IQ2_M 19.1 GiB Borderline with 8 cards. Doesn't fit
NVIDIA-Nemotron-3-Nano-30B-A3B-BF16 31.6B (3.2B active) IQ4_XS 19.1 GiB Borderline with 8 cards. Doesn't fit
MiniCPM-V-4.7-35B-A3B 34.7B (2.9B active) IQ4_XS 20.0 GiB Borderline with 8 cards. Doesn't fit
Qwen3-30B-A3B 30.5B (3.0B active) Q4_K_M 20.1 GiB Borderline with 8 cards. Doesn't fit
gemma-4-31B-it 30.7B Q3_K_M 20.3 GiB Borderline with 8 cards. Doesn't fit
EXAONE-4.5-33B 32.5B IQ4_XS 20.4 GiB Borderline with 8 cards. Doesn't fit
Llama-3.1-70B-Instruct 70.6B IQ2_XXS 22.3 GiB Borderline with 8 cards. Doesn't fit
Qwen3-Coder-Next-Base 79.7B (3.6B active) IQ2_XXS 22.3 GiB Borderline with 8 cards. Doesn't fit
Qwen3-32B 32.8B Q4_K_M 22.4 GiB Borderline with 8 cards. Doesn't fit
QwQ-32B 32.8B Q4_K_M 22.5 GiB Borderline with 8 cards. Doesn't fit
Nemotron-3-Nano-Omni-30B-A3B-Reasoning-BF16 31.6B (3.2B active) Q4_K_M 23.3 GiB Borderline with 8 cards. Doesn't fit
KAT-Dev-72B-Exp 72.7B IQ2_XXS 28.5 GiB Borderline with 8 cards. Doesn't fit
Solar-Open-100B 103B (11.2B active) IQ2_XXS 29.4 GiB Borderline with 8 cards. Doesn't fit
DeepSeek-R1-Distill-Llama-70B 70.6B Q2_K 29.4 GiB Borderline with 8 cards. Doesn't fit
Llama-3.3-70B-Instruct 70.6B Q2_K 29.4 GiB Borderline with 8 cards. Doesn't fit
Qwen3-Next-80B-A3B-Instruct 79.7B (3.6B active) Q2_K 30.3 GiB Borderline with 8 cards. Doesn't fit
Qwen2.5-72B-Instruct 73.0B Q2_K 30.3 GiB Borderline with 8 cards. Doesn't fit
Qwen3-Coder-Next 79.7B (3.6B active) Q2_K 30.4 GiB Borderline with 8 cards. Doesn't fit
GLM-4.5-Air-REAP-82B-A12B 81.9B (12.8B active) IQ2_XXS 31.1 GiB Borderline with 8 cards. Doesn't fit
Step-3.5-Flash-REAP-121B-A11B 121B (11.4B active) IQ2_XXS 32.9 GiB Borderline with 8 cards. Doesn't fit
Mistral-Small-4-119B-2603 119B (6.1B active) IQ2_XXS 34.3 GiB Borderline with 8 cards. Doesn't fit
Qwen3.5-122B-A10B 122B (9.0B active) IQ2_XXS 35.0 GiB Borderline with 8 cards. Doesn't fit
Qwen-AgentWorld-35B-A3B 34.7B (2.9B active) Q8_0 37.6 GiB Borderline with 8 cards. Doesn't fit
MiniMax-M2.5-REAP-139B-A10B 139B (10.4B active) IQ2_XXS 39.2 GiB Borderline with 8 cards. Doesn't fit
INTELLECT-3 107B (12.8B active) IQ2_XXS 39.6 GiB Borderline with 8 cards. Doesn't fit
Step-3.5-Flash-REAP-149B-A11B 149B (11.4B active) IQ2_XXS 40.1 GiB Borderline with 8 cards. Doesn't fit
Llama-4-Scout-17B-16E-Instruct 108B (16.1B active) Q2_K 41.5 GiB Borderline with 8 cards. Doesn't fit
GLM-4.5-Air 107B (12.8B active) Q2_K 45.6 GiB Borderline with 8 cards. Doesn't fit
GLM-4.6V 107B (12.8B active) Q2_K 45.6 GiB Borderline with 8 cards. Doesn't fit
Step-3.5-Flash 197B (11.5B active) IQ2_XXS 48.9 GiB Borderline with 8 cards. Doesn't fit
Devstral-2-123B-Instruct-2512 125B Q2_K 49.3 GiB Borderline with 8 cards. Doesn't fit
Mistral-Medium-3.5-128B 125B Q2_K 49.3 GiB Borderline with 8 cards. Doesn't fit
NVIDIA-Nemotron-3-Super-120B-A12B-Base-BF16 121B (12.2B active) IQ2_XXS 49.8 GiB Borderline with 8 cards. Doesn't fit
NVIDIA-Nemotron-3-Super-120B-A12B-BF16 121B (12.2B active) IQ2_XXS 51.0 GiB Borderline with 8 cards. Doesn't fit
gpt-oss-120b 117B (5.1B active) Q2_K 62.8 GiB Does not fit even on 8 cards. Doesn't fit
MiniMax-M2.7 229B (10.4B active) IQ2_XXS 62.9 GiB Does not fit even on 8 cards. Doesn't fit
Baichuan-M3-235B 235B (21.6B active) IQ2_XXS 63.2 GiB Does not fit even on 8 cards. Doesn't fit
Holo2-235B-A22B 235B (21.6B active) IQ2_XXS 63.2 GiB Does not fit even on 8 cards. Doesn't fit
Qwen3-Nemotron-235B-A22B-GenRM 235B (21.6B active) IQ2_XXS 63.2 GiB Does not fit even on 8 cards. Doesn't fit
command-a-plus-05-2026-bf16 218B (25.0B active) IQ2_XXS 63.6 GiB Does not fit even on 8 cards. Doesn't fit
NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16 121B (12.2B active) Q4_0 71.8 GiB Does not fit even on 8 cards. Doesn't fit
GLM-4.7-REAP-268B-A32B 269B (32.8B active) IQ2_XXS 73.0 GiB Does not fit even on 8 cards. Doesn't fit
Step-3.7-Flash 197B (11.5B active) IQ3_XXS 76.1 GiB Does not fit even on 8 cards. Doesn't fit
Hy3 295B (20.2B active) IQ2_XXS 82.2 GiB Does not fit even on 8 cards. Doesn't fit
MiMo-V2.5 309B (14.8B active) IQ2_XXS 82.4 GiB Does not fit even on 8 cards. Doesn't fit
GLM-4.7-REAP-218B-A32B 218B (32.8B active) Q2_K 82.9 GiB Does not fit even on 8 cards. Doesn't fit
MiniMax-M2 229B (10.4B active) Q2_K 84.9 GiB Does not fit even on 8 cards. Doesn't fit
MiniMax-M2.1 229B (10.4B active) Q2_K 84.9 GiB Does not fit even on 8 cards. Doesn't fit
MiniMax-M2.5 229B (10.4B active) Q2_K 84.9 GiB Does not fit even on 8 cards. Doesn't fit
Qwen3-235B-A22B-Instruct-2507 235B (21.6B active) Q2_K 86.7 GiB Does not fit even on 8 cards. Doesn't fit
Qwen3-VL-235B-A22B-Instruct 235B (21.6B active) Q2_K 86.7 GiB Does not fit even on 8 cards. Doesn't fit
GLM-5.3-Flash-BF16 313B (16.7B active) IQ2_XXS 92.0 GiB Does not fit even on 8 cards. Doesn't fit
Laguna-S-2.1 118B (8.1B active) Q4_K_M 95.9 GiB Does not fit even on 8 cards. Doesn't fit
Trinity-Large-Base 399B (12.8B active) IQ2_XXS 96.4 GiB Does not fit even on 8 cards. Doesn't fit
Trinity-Large-TrueBase 399B (12.8B active) IQ2_XXS 96.4 GiB Does not fit even on 8 cards. Doesn't fit
GLM-4.6-REAP-268B-A32B 269B (32.8B active) Q2_K 100.9 GiB Does not fit even on 8 cards. Doesn't fit
Trinity-Large-Thinking 399B (12.8B active) IQ2_XXS 107.1 GiB Does not fit even on 8 cards. Doesn't fit
MiMo-V2-Flash 309B (14.8B active) Q2_K 111.6 GiB Does not fit even on 8 cards. Doesn't fit
Intern-S2-397B 396B (16.3B active) IQ2_XXS 114.3 GiB Does not fit even on 8 cards. Doesn't fit
MiMo-V2.6-Flash-MOPD 309B (14.8B active) Q2_K 124.9 GiB Does not fit even on 8 cards. Doesn't fit
MiMo-V2.6-Flash-RL 309B (14.8B active) Q2_K 124.9 GiB Does not fit even on 8 cards. Doesn't fit
K-EXAONE-236B-A23B 237B (22.7B active) IQ4_XS 126.3 GiB Does not fit even on 8 cards. Doesn't fit
GLM-4.5 353B (32.9B active) Q2_K 130.7 GiB Does not fit even on 8 cards. Doesn't fit
GLM-4.7 353B (32.9B active) Q2_K 130.7 GiB Does not fit even on 8 cards. Doesn't fit
Qwen3-Coder-REAP-363B-A35B 363B (34.5B active) Q2_K 132.7 GiB Does not fit even on 8 cards. Doesn't fit
Qwen3-235B-A22B 235B (21.6B active) Q4_K_M 141.9 GiB Does not fit even on 8 cards. Doesn't fit
Qwen3-Coder-480B-A35B-Instruct 480B (34.5B active) Q2_K 174.3 GiB Does not fit even on 8 cards. Doesn't fit
Qwen3.5-397B-A17B 396B (16.3B active) Q3_K_M 175.6 GiB Does not fit even on 8 cards. Doesn't fit
GLM-5.1 744B (40.8B active) IQ2_XXS 197.6 GiB Does not fit even on 8 cards. Doesn't fit
DeepSeek-V3.2 671B (36.6B active) Q2_K 242.1 GiB Does not fit even on 8 cards. Doesn't fit
cogito-671b-v2.1 671B (36.6B active) Q2_K 242.4 GiB Does not fit even on 8 cards. Doesn't fit
DeepSeek-V3.1 671B (36.6B active) Q2_K 242.4 GiB Does not fit even on 8 cards. Doesn't fit
Mistral-Large-3-675B-Instruct-2512 673B (39.0B active) Q2_K 243.8 GiB Does not fit even on 8 cards. Doesn't fit
DeepSeek-V3 671B (36.6B active) Q2_K 257.8 GiB Does not fit even on 8 cards. Doesn't fit
Kimi-K2.6 1.03T (31.7B active) IQ2_XXS 267.6 GiB Does not fit even on 8 cards. Doesn't fit
GLM-5 744B (40.8B active) Q2_K 268.9 GiB Does not fit even on 8 cards. Doesn't fit
MiMo-V2.5-Pro 1.02T (41.0B active) IQ2_XXS 270.4 GiB Does not fit even on 8 cards. Doesn't fit
DeepSeek-R1 671B (36.6B active) Q2_K 280.3 GiB Does not fit even on 8 cards. Doesn't fit
Muse-Spark-Open-816B-A42B 820B (44.4B active) Q2_K 301.0 GiB Does not fit even on 8 cards. Doesn't fit
GLM-5.3-Flash 313B (16.7B active) Q8_0 327.9 GiB Does not fit even on 8 cards. Doesn't fit
Ling-1T 1.00T (49.8B active) Q2_K 360.6 GiB Does not fit even on 8 cards. Doesn't fit
Kimi-K2-Instruct 1.03T (31.7B active) Q2_K 367.4 GiB Does not fit even on 8 cards. Doesn't fit
Kimi-K2.5 1.03T (31.7B active) Q2_K 368.0 GiB Does not fit even on 8 cards. Doesn't fit
Kimi-K2-Thinking 1.03T (31.7B active) Q2_K 368.3 GiB Does not fit even on 8 cards. Doesn't fit
MiMo-V2.6-Pro-MOPD 1.02T (41.0B active) Q2_K 411.6 GiB Does not fit even on 8 cards. Doesn't fit
MiMo-V2.6-Pro-RL 1.02T (41.0B active) Q2_K 411.6 GiB Does not fit even on 8 cards. Doesn't fit
Beam-1.0-501B-A23B 501B (22.7B active) Q8_0 523.6 GiB Does not fit even on 8 cards. Doesn't fit
GLM-5.2 744B (40.8B active) Q8_0 776.4 GiB Does not fit even on 8 cards. Doesn't fit
GLM-5.3 744B (40.8B active) Q8_0 776.4 GiB Does not fit even on 8 cards. Doesn't fit
Qwen3.8-2.4T-A95B 2.42T (93.3B active) Q8_0 2,524.1 GiB Does not fit even on 8 cards. Doesn't fit

The speed ceiling is memory bandwidth divided by the bytes read per generated token. It is a theoretical upper bound; public llama.cpp benchmarks typically reach about half to four-fifths of it, and small models and mixture-of-experts models often less. Public benchmarks

Other devices with 6 GB