Your rig

H100 PCIe · 80 GB

16 GB system RAM · Hopper · 2000 GB/s · 78.4 GB usable VRAM
Edit hardware Export as JSON

Context length

KV cache at f16, 4K tokens

Quality floor
Q8_0 — near-lossless Q6_K Q5_K_M Q4_K_M — recommended Q3_K_M Q2_K — damaged
KV cache type
System RAM (GB)
Runtime
Ollamallama.cppLM StudiovLLM

4 of 5 runtimes drive this device.

Allow CPU offload Include partial-offload fits

Offloaded layers run at 60 GB/s host bandwidth — expect single-digit tokens per second.

Which one should I actually run?

Best models for your H100 PCIe

Updated 8 Oct 2026 1 new this month

What do you want to do? Each row is the best-ranked model on that use case's shortlist that fits at Q3_K_M and 4K. The ordering is editorial; the sizes and speeds are computed. How the shortlists work.

RecommendationModelWhyQuant · totalTok/s est.
Best overall Qwen3.8 27B
Alibaba · Apache 2.0
Best capability per gigabyte: a current-generation dense 27B with vision and 262K context. Q3_K_M · 13.5 GB 96 Runs great
Best for coding Ling 3.0 Flash 124B-A5B
inclusionAI · MIT
SWE-bench Pro 56.6 claimed at 5B active — the 128 GB-class coding pick if you cannot fit DeepSeek V4. Q3_K_M · 57.1 GB 201 Runs great
Best reasoning Ling 3.0 Flash 124B-A5B
inclusionAI · MIT
AIME 93 claimed at 5B active; hybrid attention keeps long reasoning traces cheap. Q3_K_M · 57.1 GB 201 Runs great
Best for writing Llama 3.3 70B Instruct
Meta · Llama 3.3 Community
The creative-writing favourite: consistent voice, takes direction, handles fiction framing. Q3_K_M · 34.0 GB 38 Runs great
Best vision Qwen3.8 27B
Alibaba · Apache 2.0
Image and video in, OSWorld 84 — the strongest local vision-language model at 24 GB. Q3_K_M · 13.5 GB 96 Runs great
Best for agents Ling 3.0 Flash 124B-A5B
inclusionAI · MIT
Agentic workflows are what Ling 3.0 was trained for; MIT, 5B active. Q3_K_M · 57.1 GB 201 Runs great
Best translation Qwen3.8 27B
Alibaba · Apache 2.0
119 languages, and Qwen translations read as idiomatic rather than literal. Q3_K_M · 13.5 GB 96 Runs great
Fastest good model Qwen3.6 35B-A3B
Alibaba · Apache 2.0
3.3B active per token: the fastest model that still competes with dense 27Bs. Q3_K_M · 17.0 GB 310 Runs great
Best long context Qwen3.8 27B
Alibaba · Apache 2.0
262K native, and only 16 of 64 blocks keep a KV cache — long context is cheap here. Q3_K_M · 21.3 GB
at 128K context
96 Runs great

Everything that fits

Runs great

66 of 80 · fits with headroom for a longer prompt
ModelParamsQuantWeights +KV 4KTotal VRAM headroom Tok/sRuntime
Qwen3.5 122B-A10B 125B MoE / 10B act Q3_K_M 56.9 GB 0.09 GB 57.6 GB
20.8 GB free
102 Ollama Run
Nemotron 3 Super 120B-A12B 124B MoE / 12B act Q3_K_M 56.4 GB 0.03 GB 57.1 GB
21.3 GB free
85 llama.cpp Run
Ling 3.0 Flash 124B-A5B 124B MoE / 5.1B act Q3_K_M 56.4 GB 0.03 GB 57.1 GB
21.3 GB free
201 llama.cpp Run
Mistral Small 4 119B-A6B 119B MoE / 6.5B act Q3_K_M 54.2 GB 0.09 GB 54.9 GB
23.5 GB free
157 llama.cpp Run
gpt-oss 120B 117B MoE / 5.1B act MXFP4 60.5 GB 0.15 GB 61.3 GB
17.1 GB free
176 Ollama Run
Llama 4 Scout 109B-A17B 109B MoE / 17B act Q3_K_M 49.6 GB 0.75 GB 51.0 GB
27.4 GB free
60 Ollama Run
Llama 3.3 70B Instruct 70.6B Q3_K_M 32.1 GB 1.25 GB 34.0 GB
44.4 GB free
38 Ollama Run
DeepSeek-R1-Distill-Llama 70B 70.6B Q3_K_M 32.1 GB 1.25 GB 34.0 GB
44.4 GB free
38 Ollama Run
Qwen3.6 35B-A3B 35.9B MoE / 3.3B act Q3_K_M 16.3 GB 0.08 GB 17.0 GB
61.4 GB free
310 Ollama Run
Ornith 1.5 35B-A3B 35.9B MoE / 3.3B act Q3_K_M 16.3 GB 0.08 GB 17.0 GB
61.4 GB free
310 Ollama Run
LLM-jp 4 33B Thinking 33.2B Q3_K_M 15.1 GB 1.00 GB 16.7 GB
61.7 GB free
80 llama.cpp Run
Qwen3 32B 32.8B Q3_K_M 14.9 GB 1.00 GB 16.5 GB
61.9 GB free
81 Ollama Run
DeepSeek-R1-Distill-Qwen 32B 32.8B Q3_K_M 14.9 GB 1.00 GB 16.5 GB
61.9 GB free
81 Ollama Run
Qwen2.5-Coder 32B 32.8B Q3_K_M 14.9 GB 1.00 GB 16.5 GB
61.9 GB free
81 Ollama Run
Olmo 3.1 32B Instruct 32.2B Q3_K_M 14.7 GB 1.00 GB 16.3 GB
62.1 GB free
83 Ollama Run
Nemotron 3.5 Lightning 30B-A3B 31.6B MoE / 3.2B act Q3_K_M 14.4 GB 0.02 GB 15.0 GB
63.4 GB free
320 Ollama Run
Gemma 4 31B 31.3B Q3_K_M 14.2 GB 1.41 GB 16.3 GB
62.1 GB free
85 Ollama Run
GLM-4.7-Flash 30B-A3B 31.2B MoE / 3B act Q3_K_M 14.2 GB 0.21 GB 15.0 GB
63.4 GB free
341 Ollama Run
Qwen3 30B-A3B 30.5B MoE / 3.3B act Q3_K_M 13.9 GB 0.38 GB 14.9 GB
63.5 GB free
310 Ollama Run
Qwen3 Coder 30B-A3B 30.5B MoE / 3.3B act Q3_K_M 13.9 GB 0.38 GB 14.9 GB
63.5 GB free
310 Ollama Run
Muse Glimmer 30B 29.8B Q3_K_M 13.6 GB 0.13 GB 14.3 GB
64.1 GB free
89 Ollama Run
Granite 4.1 30B 28.9B Q3_K_M 13.2 GB 1.00 GB 14.8 GB
63.6 GB free
92 Ollama Run
Qwen3.8 27B 27.8B Q3_K_M 12.7 GB 0.25 GB 13.5 GB
64.9 GB free
96 Ollama Run
Qwen3.6 27B 27.8B Q3_K_M 12.7 GB 0.25 GB 13.5 GB
64.9 GB free
96 Ollama Run
Gemma 3 27B 27.4B Q3_K_M 12.5 GB 0.72 GB 13.8 GB
64.6 GB free
97 Ollama Run
Gemma 4 26B-A4B 26.5B MoE / 3.8B act Q3_K_M 12.1 GB 0.35 GB 13.0 GB
65.4 GB free
269 Ollama Run
Devstral Small 2 24B 24B Q3_K_M 10.9 GB 0.63 GB 12.1 GB
66.3 GB free
111 llama.cpp Run
Mistral Small 3.2 24B 23.6B Q3_K_M 10.7 GB 0.63 GB 12.0 GB
66.4 GB free
113 Ollama Run
gpt-oss 20B 20.9B MoE / 3.6B act MXFP4 10.8 GB 0.10 GB 11.5 GB
66.9 GB free
250 Ollama Run
Qwen3 14B 14.8B Q3_K_M 6.7 GB 0.63 GB 8.0 GB
70.4 GB free
180 Ollama Run
DeepSeek-R1-Distill-Qwen 14B 14.8B Q3_K_M 6.7 GB 0.75 GB 8.1 GB
70.3 GB free
180 Ollama Run
Qwen2.5-Coder 14B 14.8B Q3_K_M 6.7 GB 0.75 GB 8.1 GB
70.3 GB free
180 Ollama Run
Phi-4 14B 14.7B Q3_K_M 6.7 GB 0.78 GB 8.1 GB
70.3 GB free
181 Ollama Run
Ministral 3 14B 13.9B Q3_K_M 6.3 GB 0.63 GB 7.6 GB
70.8 GB free
191 Ollama Run
Gemma 3 12B 12.2B Q3_K_M 5.6 GB 0.56 GB 6.7 GB
71.7 GB free
218 Ollama Run
Mistral NeMo 12B 12.2B Q3_K_M 5.6 GB 0.63 GB 6.8 GB
71.6 GB free
218 Ollama Run
Gemma 4 12B 12B Q3_K_M 5.5 GB 0.56 GB 6.6 GB
71.8 GB free
222 Ollama Run
Qwen3.5 9B 9.65B Q3_K_M 4.4 GB 0.13 GB 5.1 GB
73.3 GB free
276 Ollama Run
Ornith 1.5 9B 9.41B Q3_K_M 4.3 GB 0.13 GB 5.0 GB
73.4 GB free
283 Ollama Run
Ministral 3 8B 8.92B Q3_K_M 4.1 GB 0.53 GB 5.2 GB
73.2 GB free
298 Ollama Run
Granite 4.1 8B 8.79B Q3_K_M 4.0 GB 0.63 GB 5.2 GB
73.2 GB free
303 Ollama Run
LFM2.5 8B-A1B 8.47B MoE / 1.5B act Q3_K_M 3.9 GB 0.05 GB 4.5 GB
73.9 GB free
682 Ollama Run
Fara 7B 8.29B Q3_K_M 3.8 GB 0.22 GB 4.6 GB
73.8 GB free
321 llama.cpp Run
Qwen3 8B 8.19B Q3_K_M 3.7 GB 0.56 GB 4.9 GB
73.5 GB free
325 Ollama Run
Llama 3.1 8B Instruct 8.03B Q3_K_M 3.7 GB 0.50 GB 4.8 GB
73.6 GB free
331 Ollama Run
Gemma 4 E4B 8.0B Q3_K_M 3.6 GB 0.09 GB 4.3 GB
74.1 GB free
332 Ollama Run
Ling 3.0 Tiny 7.9B-A1.3B 7.9B MoE / 1.3B act Q3_K_M 3.6 GB 0.03 GB 4.2 GB
74.2 GB free
787 llama.cpp Run
DeepSeek-R1-Distill-Qwen 7B 7.62B Q3_K_M 3.5 GB 0.22 GB 4.3 GB
74.1 GB free
349 Ollama Run
Qwen2.5-Coder 7B 7.62B Q3_K_M 3.5 GB 0.22 GB 4.3 GB
74.1 GB free
349 Ollama Run
Olmo 3 7B Instruct 7.3B Q3_K_M 3.3 GB 2.00 GB 5.9 GB
72.5 GB free
364 Ollama Run
Mistral 7B Instruct v0.3 7.25B Q3_K_M 3.3 GB 0.50 GB 4.4 GB
74.0 GB free
367 Ollama Run
Gemma 4 E2B 5.1B Q3_K_M 2.3 GB 0.04 GB 3.0 GB
75.4 GB free
522 Ollama Run
Qwen3.5 4B 4.66B Q3_K_M 2.1 GB 0.13 GB 2.8 GB
75.6 GB free
571 Ollama Run
Gemma 3 4B 4.3B Q3_K_M 2.0 GB 0.20 GB 2.8 GB
75.6 GB free
619 Ollama Run
Qwen3 4B 4.02B Q3_K_M 1.8 GB 0.56 GB 3.0 GB
75.4 GB free
662 Ollama Run
Ministral 3 3B 3.85B Q3_K_M 1.8 GB 0.41 GB 2.8 GB
75.6 GB free
691 Ollama Run
Phi-4-mini 3.8B 3.84B Q3_K_M 1.7 GB 0.50 GB 2.8 GB
75.6 GB free
693 Ollama Run
Granite 4.1 3B 3.4B Q3_K_M 1.5 GB 0.31 GB 2.5 GB
75.9 GB free
782 Ollama Run
Llama 3.2 3B Instruct 3.21B Q3_K_M 1.5 GB 0.44 GB 2.5 GB
75.9 GB free
829 Ollama Run
LFM2.5 2.6B 2.7B Q3_K_M 1.2 GB 0.06 GB 1.9 GB
76.5 GB free
985 llama.cpp Run
Qwen3.5 2B 2.27B Q3_K_M 1.0 GB 0.05 GB 1.7 GB
76.7 GB free
1172 Ollama Run
Qwen3 1.7B 1.72B Q3_K_M 0.8 GB 0.44 GB 1.8 GB
76.6 GB free
1546 Ollama Run
Llama 3.2 1B Instruct 1.24B Q3_K_M 0.6 GB 0.13 GB 1.3 GB
77.1 GB free
2145 Ollama Run
Gemma 3 1B 1.0B Q3_K_M 0.5 GB 0.03 GB 1.1 GB
77.3 GB free
2660 Ollama Run
Qwen3.5 0.8B 0.87B Q3_K_M 0.4 GB 0.05 GB 1.0 GB
77.4 GB free
3057 Ollama Run
Qwen3 0.6B 0.6B Q3_K_M 0.3 GB 0.44 GB 1.3 GB
77.1 GB free
4433 Ollama Run

Won't fit in VRAM

14 models · some are possible with CPU offload
ModelParamsQuantWeights +KV 4KTotal Over budget Tok/sRuntime
Qwen3.8-Flash-Next 180B-A6B 180B MoE / 6B act Q3_K_M 81.9 GB 0.09 GB 82.6 GB +4.2 GB ~64 llama.cpp Why
Kimi K3 2.8T-A104B 2780B MoE / 104B act Q3_K_M 1265.4 GB 0.11 GB 1266.1 GB +1187.7 GB ~0.3 llama.cpp Why
Qwen3.8 2.4T-A95B 2446B MoE / 95B act Q3_K_M 1113.4 GB 0.36 GB 1114.3 GB +1035.9 GB ~0.3 llama.cpp Why
DeepSeek V4 Pro 1.6T-A49B 1650B MoE / 49B act Q3_K_M 751.1 GB 0.27 GB 751.9 GB +673.5 GB ~0.7 llama.cpp Why
Kimi K2.6 1T-A32B 1027B MoE / 32B act Q3_K_M 467.5 GB 0.27 GB 468.3 GB +389.9 GB ~1.1 llama.cpp Why
GLM-5.2 744B-A40B 753B MoE / 40B act Q3_K_M 342.8 GB 0.34 GB 343.7 GB +265.3 GB ~1.0 llama.cpp Why
GLM-5.3 744B-A40B 753B MoE / 40B act Q3_K_M 342.8 GB 0.34 GB 343.7 GB +265.3 GB ~1.0 llama.cpp Why
DeepSeek-R1 671B 671B MoE / 37B act Q3_K_M 305.4 GB 0.27 GB 306.3 GB +227.9 GB ~1.1 llama.cpp Why
DeepSeek V4.1 Flash 552B-A16B New 552B MoE / 16B act Q3_K_M 251.3 GB 0.02 GB 251.9 GB +173.5 GB ~2.7 vLLM Why
Qwen3.5 397B-A17B 403B MoE / 17B act Q3_K_M 183.4 GB 0.12 GB 184.2 GB +105.8 GB ~3.1 llama.cpp Why
Ornith 1.5 397B-A17B 397B MoE / 17B act Q3_K_M 180.7 GB 0.12 GB 181.4 GB +103.0 GB ~3.1 llama.cpp Why
GLM-5.3-Flash 320B-A18B 321B MoE / 18B act Q3_K_M 146.1 GB 0.04 GB 146.8 GB +68.4 GB ~3.5 llama.cpp Why
DeepSeek V4 Flash 284B-A13B 284B MoE / 13B act Q3_K_M 129.3 GB 0.19 GB 130.1 GB +51.7 GB ~5.7 llama.cpp Why
Qwen3 235B-A22B 235B MoE / 22B act Q3_K_M 107.0 GB 0.73 GB 108.3 GB +29.9 GB ~4.6 llama.cpp Why

Totals = quantised weights + f16 KV cache at 4K tokens + 0.6 GB runtime overhead, against 78.4 GB usable VRAM (80 GB less display and driver reserve). Tokens per second are estimated from 2000 GB/s peak memory bandwidth at batch 1 — not measured. Read the method.


Same memory, different speed

These devices have the same 78.4 GB usable, so they run exactly the models above. Generation speed scales with memory bandwidth.

DeviceBandwidthSpeed vs yours
H100 SXM
Hopper · 2022
3350 GB/s 1.68×
A100 80 GB
Ampere · 2021
2039 GB/s 1.02×