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.

inclusionAI logo

Can the RTX 3090 run ZwZ2-27B?

ZwZ2-27B·27.4B·GeForce RTX 3090·24 GB VRAM

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

Yes. The RTX 3090 runs ZwZ2-27B fully in VRAM at Q5_K_M (about 21 GB of 24 GB), at roughly 26 tokens/s.

best quant
Q5_K_M
needs
21 GB
usable
24 GB
est. speed
~26 tok/s

Estimates for an 8K context; offload assumes 32 GB of system RAM.Try other contexts on the ZwZ2-27B page →

Every quantisation of ZwZ2-27B on the RTX 3090

QuantQualityMemoryFitSpeed
Q2_K Heavy loss11 GBfits~57 tok/s
Q3_K_M Noticeable loss16 GBfits~39 tok/s
Q4_K_M Small loss19 GBfits~31 tok/s
Q5_K_M pickNear-lossless21 GBfits~26 tok/s
Q6_K Near-lossless24 GBoffload~22 tok/s
Q8_0 Lossless30 GBoffload~4.3 tok/s
FP16 / BF16 Lossless55 GBtoo big—

Speed is a bandwidth-based estimate for generating text (936 GB/s); prompt processing and long contexts are slower.

Measured speed on the RTX 3090

No community reports for this pair yet.

Report your own speed →

Run ZwZ2-27B on the RTX 3090

Using Q5_K_M, the best fit for RTX 3090.

llama.cpp · OpenAI-compatible server with a web UIdocs
llama-server -hf prithivMLmods/inclusionAI-ZwZ2-27B-GGUF:Q5_K_M

Downloads the GGUF on first run. Installed through llama.app? The same flags work with `llama serve`.

vLLM · Serve an OpenAI-compatible APIdocs
vllm serve inclusionAI/ZwZ2-27B
vLLM · Docker Compose (NVIDIA)docs
services:
  vllm:
    image: vllm/vllm-openai:latest
    command: inclusionAI/ZwZ2-27B
    ports:
      - "8000:8000"
    ipc: host
    volumes:
      - ~/.cache/huggingface:/root/.cache/huggingface
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]

API on port 8000.

SGLang · Serve an OpenAI-compatible APIdocs
sglang serve inclusionAI/ZwZ2-27B --host 0.0.0.0 --port 30000
SGLang · Docker Compose (NVIDIA)docs
services:
  sglang:
    image: lmsysorg/sglang:latest
    command: sglang serve inclusionAI/ZwZ2-27B --host 0.0.0.0 --port 30000
    ports:
      - "30000:30000"
    ipc: host
    volumes:
      - ~/.cache/huggingface:/root/.cache/huggingface
    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="inclusionAI/ZwZ2-27B", 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"])

Vision models that fit the RTX 3090 comfortably

All that fit →
ModelParamsBest quantMemorySpeedLicence
27.8BQ5_K_M≈22 GBFits in memory~26 tok/sCommercial OK
27.8BQ5_K_M≈22 GBFits in memory~26 tok/sCommercial OK
25.8BA4BQ5_K_M≈20 GBFits in memory~181 tok/sCommercial OK
27.4BQ4_K_M≈21 GBFits in memory~31 tok/sConditions
27.8BQ5_K_M≈22 GBFits in memory~26 tok/sCommercial OK
meta-models logoMuse-Glimmer-30Bmeta-models
29.8BQ5_K_M≈22 GBFits in memory~24 tok/sCommercial OK

Other GPUs that run ZwZ2-27B

Frequently asked questions

Can the RTX 3090 run ZwZ2-27B?
Yes. The RTX 3090 runs ZwZ2-27B fully in VRAM at Q5_K_M (about 21 GB of 24 GB), at roughly 26 tokens/s.
How fast does ZwZ2-27B run on the RTX 3090?
An estimate from memory bandwidth gives about 26 tokens/s at Q5_K_M.
Which quantisation of ZwZ2-27B should I use on the RTX 3090?
Q5_K_M: the largest one that fits. Expected quality: near-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.