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.

microsoft logo

Can the Radeon RX 6700 XT run UserLM-2.0?

UserLM-2.0·8B·Radeon RX 6700 XT·12 GB VRAM

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

Yes. The Radeon RX 6700 XT runs UserLM-2.0 fully in VRAM at Q8_0 (about 9.8 GB of 12 GB), at roughly 25 tokens/s.

best quant
Q8_0
needs
9.8 GB
usable
12 GB
est. speed
~25 tok/s

Estimates for an 8K context; offload assumes 32 GB of system RAM.Try other contexts on the UserLM-2.0 page →

Every quantisation of UserLM-2.0 on the Radeon RX 6700 XT

QuantQualityMemoryFitSpeed
Q2_K Heavy loss4.1 GBfits~80 tok/s
Q3_K_M Noticeable loss5.3 GBfits~54 tok/s
Q4_K_M Small loss6.3 GBfits~43 tok/s
Q5_K_M Near-lossless7.1 GBfits~37 tok/s
Q6_K Near-lossless8.0 GBfits~32 tok/s
Q8_0 pickLossless9.8 GBfits~25 tok/s
FP16 / BF16 Lossless17 GBoffload~5 tok/s

Speed is a bandwidth-based estimate for generating text (384 GB/s); prompt processing and long contexts are slower.

Measured speed on the Radeon RX 6700 XT

No community reports for this pair yet.

Report your own speed →

Run UserLM-2.0 on the Radeon RX 6700 XT

Using Q8_0, the best fit for Radeon RX 6700 XT.

llama.cpp · OpenAI-compatible server with a web UIdocs
llama-server -hf liodon-ai/UserLM-2.0-imatrix-GGUF:Q8_0

Needs a HIP (ROCm) build of llama.cpp. Start it with HSA_OVERRIDE_GFX_VERSION=10.3.0 in front for this card. Downloads the GGUF on first run.

llama.cpp · Docker with ROCm (AMD)docs
docker run -d --name llama --device /dev/kfd --device /dev/dri -p 8080:8080 \
  -e HSA_OVERRIDE_GFX_VERSION=10.3.0 \
  -v ~/.cache/llama.cpp:/root/.cache/llama.cpp \
  ghcr.io/ggml-org/llama.cpp:server-rocm \
  -hf liodon-ai/UserLM-2.0-imatrix-GGUF:Q8_0 --host 0.0.0.0 --port 8080 -ngl 99

Server and web UI on port 8080.

vLLM · Serve an OpenAI-compatible APIdocs
vllm serve microsoft/UserLM-2.0

Needs vLLM built for ROCm; it supports Instinct MI200/MI300 and Radeon RX 7900 / 9000 cards.

vLLM · Smaller FP8 builddocs
vllm serve liodon-ai/UserLM-2.0-FP8
vLLM · Docker Compose (AMD ROCm)docs
services:
  vllm:
    image: rocm/vllm:latest
    command: vllm serve microsoft/UserLM-2.0 --host 0.0.0.0 --port 8000
    ports:
      - "8000:8000"
    ipc: host
    devices:
      - /dev/kfd
      - /dev/dri
    group_add:
      - video
    security_opt:
      - seccomp=unconfined
    volumes:
      - ~/.cache/huggingface:/root/.cache/huggingface

AMD’s rocm/vllm image. API on port 8000.

SGLang · Serve an OpenAI-compatible APIdocs
sglang serve microsoft/UserLM-2.0 --host 0.0.0.0 --port 30000

Needs SGLang’s ROCm build (Instinct MI300-class GPUs).

Transformers · Chat with the pipeline APIdocs
from transformers import pipeline

pipe = pipeline("text-generation", model="microsoft/UserLM-2.0", dtype="auto", device_map="auto")
messages = [{"role": "user", "content": "Explain self-hosting in one paragraph."}]
out = pipe(messages, max_new_tokens=512)
print(out[0]["generated_text"][-1]["content"])

Chat models that fit the Radeon RX 6700 XT comfortably

All that fit →
ModelParamsBest quantMemorySpeedLicence
8BQ8_0≈9.8 GBFits in memory~25 tok/sConditions
8.2BQ8_0≈10 GBFits in memory~24 tok/sCommercial OK
8BQ8_0≈9.8 GBFits in memory~25 tok/sConditions
7.6BQ8_0≈8.9 GBFits in memory~26 tok/sCommercial OK
9.4BQ8_0≈11 GBFits in memory~21 tok/sCommercial OK
7.2BQ8_0≈9.0 GBFits in memory~27 tok/sCommercial OK

Other GPUs that run UserLM-2.0

Frequently asked questions

Can the Radeon RX 6700 XT run UserLM-2.0?
Yes. The Radeon RX 6700 XT runs UserLM-2.0 fully in VRAM at Q8_0 (about 9.8 GB of 12 GB), at roughly 25 tokens/s.
How fast does UserLM-2.0 run on the Radeon RX 6700 XT?
An estimate from memory bandwidth gives about 25 tokens/s at Q8_0.
Which quantisation of UserLM-2.0 should I use on the Radeon RX 6700 XT?
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.