selfhost.directory

The index of self-hostable projects — ranked, categorized and served with live updates.
Own your data and stop giving it to big tech.

h2oai logo

Can the RTX A6000 run h2o-lightning-31b?

h2o-lightning-31b·31.3B·RTX A6000·48 GB VRAM

Saved in this browser, used on every model page.
Yes

Yes. The RTX A6000 runs h2o-lightning-31b fully in VRAM at Q8_0 (about 35 GB of 48 GB), at roughly 13 tokens/s.

best quant
Q8_0
needs
35 GB
usable
48 GB
est. speed
~13 tok/s

Estimates for an 8K context; offload assumes 32 GB of system RAM.Try other contexts on the h2o-lightning-31b page →

Every quantisation of h2o-lightning-31b on the RTX A6000

QuantQualityMemoryFitSpeed
Q2_K Heavy loss13 GBfits~41 tok/s
Q3_K_M Noticeable loss17 GBfits~28 tok/s
Q4_K_M Small loss21 GBfits~22 tok/s
Q5_K_M Near-lossless24 GBfits~19 tok/s
Q6_K Near-lossless28 GBfits~16 tok/s
Q8_0 pickLossless35 GBfits~13 tok/s
FP16 / BF16 Lossless64 GBoffload~1.7 tok/s

Speed is a bandwidth-based estimate for generating text (768 GB/s); prompt processing and long contexts are slower.

Measured speed on the RTX A6000

No community reports for this pair yet.

Report your own speed →

Run h2o-lightning-31b on the RTX A6000

vLLM · Serve an OpenAI-compatible APIdocs
vllm serve h2oai/h2o-lightning-31b

Gated on Hugging Face: accept the licence on the model page and export HF_TOKEN first.

vLLM · Docker Compose (NVIDIA)docs
services:
  vllm:
    image: vllm/vllm-openai:latest
    command: h2oai/h2o-lightning-31b
    ports:
      - "8000:8000"
    ipc: host
    volumes:
      - ~/.cache/huggingface:/root/.cache/huggingface
    environment:
      - HF_TOKEN=${HF_TOKEN}
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]

API on port 8000.

SGLang · Serve an OpenAI-compatible APIdocs
sglang serve h2oai/h2o-lightning-31b --host 0.0.0.0 --port 30000

Gated on Hugging Face: accept the licence on the model page and export HF_TOKEN first.

SGLang · Docker Compose (NVIDIA)docs
services:
  sglang:
    image: lmsysorg/sglang:latest
    command: sglang serve h2oai/h2o-lightning-31b --host 0.0.0.0 --port 30000
    ports:
      - "30000:30000"
    ipc: host
    volumes:
      - ~/.cache/huggingface:/root/.cache/huggingface
    environment:
      - HF_TOKEN=${HF_TOKEN}
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]

API on port 30000.

Transformers · Ask about an imagedocs
from transformers import pipeline

pipe = pipeline("image-text-to-text", model="h2oai/h2o-lightning-31b", dtype="auto", device_map="auto")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/bee.jpg"},
            {"type": "text", "text": "Describe this image."},
        ],
    }
]
out = pipe(text=messages, max_new_tokens=256)
print(out[0]["generated_text"][-1]["content"])

Gated on Hugging Face: accept the licence on the model page and export HF_TOKEN first.

Vision models that fit the RTX A6000 comfortably

All that fit →
ModelParamsBest quantMemorySpeedLicence
31.3BQ8_0≈40 GBFits in memory~13 tok/sCommercial OK
27.8BQ8_0≈31 GBFits in memory~14 tok/sCommercial OK
36BA3BQ8_0≈38 GBFits in memory~133 tok/sCommercial OK
meta-models logoMuse-Glimmer-30Bmeta-models
29.8BQ8_0≈30 GBFits in memory~13 tok/sCommercial OK
27.8BQ8_0≈30 GBFits in memory~14 tok/sCommercial OK
25.8BA4BQ8_0≈29 GBFits in memory~99 tok/sCommercial OK

Other GPUs that run h2o-lightning-31b

Frequently asked questions

Can the RTX A6000 run h2o-lightning-31b?
Yes. The RTX A6000 runs h2o-lightning-31b fully in VRAM at Q8_0 (about 35 GB of 48 GB), at roughly 13 tokens/s.
How fast does h2o-lightning-31b run on the RTX A6000?
An estimate from memory bandwidth gives about 13 tokens/s at Q8_0.
Which quantisation of h2o-lightning-31b should I use on the RTX A6000?
Q8_0: the largest one that fits. Expected quality: lossless compared with the full-precision weights.

Memory and speed are estimates from the model's size and the card's memory bandwidth; real numbers depend on the runtime, drivers and context length. Community figures are medians of user reports.