VOOZH about

URL: https://willitrunai.com/can-run/cogvlm2-19b-on-radeon-pro-w6800-32gb

⇱ CogVLM2 19B on Radeon Pro W6800 32GB? YES

Can CogVLM2 19B run on Radeon Pro W6800 32GB?

YES — Runs Great

A83Great
○Estimated from fit model

CogVLM2 19B needs ~18.1 GB VRAM. Radeon Pro W6800 32GB has 32.0 GB. With Q4_K_M quantization, expect ~27 tok/s.

Runtime: llama.cppCapacity: RoomyBandwidth: MediumStack: StandardBottleneck: Balanced
Share:

Operating mode

Choose the run profile you care about

Interactive favors responsiveness, while light API and scale-out lean harder on serving readiness. The fit stays the same, but the recommendation lens changes.

Current mode

Balanced

Balanced for general local use. Keeps the ranking neutral across personal and serving workflows.

Capabilities:

Select quantization to explore

Q4_K_M (Medium quality) — 18.1 GB, 26.6 tok/s, Runs well
18.1 GB required32.0 GB available
57% VRAM used

Fit status

Runs well

Decode

26.6 tok/s

TTFT

7280 ms

Safe context

8K

Memory

18.1 GB / 32.0 GB

Memory breakdown

Weights11.6 GB
KV Cache2.4 GB
Runtime0.9 GB
Headroom3.2 GB

See how fast it feels

See how fast it feelsCogVLM2 19B on Radeon Pro W6800 32GB
1st promptCold start — includes initialization
>What is local AI inference?
Local AI inference means running an AI language model directly on your own hardware — your laptop, desktop, or server — instead of sending requests to a remote cloud API. When you run inference locally the model weights are loaded into your GPU or unified memory. Each token you generate requires reading those weights from memory, so memory bandwidth is the main bottleneck for decode speed. Key benefits of running locally: - Full privacy: your prompts never leave your machine - No per-token cost or rate limits - Works offline once the model is downloaded - Latency depends only on your hardware
2nd promptWarm — model ready, no init overhead
>How much VRAM do I need?
It depends on the model size and quantization level. A rough rule of thumb: Model size Q4 (4-bit) Q8 (8-bit) FP16 7B params ~4.3 GB ~7.5 GB ~14 GB 13B params ~7.9 GB ~13.9 GB ~26 GB 70B params ~42.7 GB ~74.9 GB ~140 GB Most people use 4-bit quantization (Q4_K_M) which gives 90-95% of full quality at a fraction of the memory. A 24 GB GPU can comfortably run most 7B-13B models.
Estimated: 26.6 tok/s decode · 7.3s TTFT (warm) · 67 tok/s prefill

What limits this setup

This setup is broadly balanced for this model.

No major red flags

This recommendation has enough memory headroom and acceptable estimated speed for the selected workload.

Best improvement path

Performance by workload

WorkloadGradeFitDecodeTTFTContext
ChatARuns well26.6 tok/s3971 ms8K
CodingARuns well26.6 tok/s7280 ms8K
Agentic CodingARuns well26.6 tok/s10589 ms8K
ReasoningARuns well26.6 tok/s8603 ms8K
RAGARuns well26.6 tok/s13236 ms8K

Quantization options

How CogVLM2 19B (19B params) fits at each quantization level on Radeon Pro W6800 32GB (32.0 GB usable).

QuantBitsVRAMQualityFit
Q2_K
2
7.4 GB
LowA78
Q3_K_S
3
9.3 GB
LowA79
NVFP4
4
10.6 GB
MediumA79
Q4_K_M
4
11.6 GB
MediumA80
Q5_K_M
5
13.7 GB
HighA81
Q6_K
6
15.6 GB
HighA82
Q8_0Best for your GPU
8
20.3 GB
Very HighA82
F16
16
38.9 GB
MaximumF0

Get started

Copy-paste commands to run CogVLM2 19B on your machine.

Run

docker run --rm -it ghcr.io/ggerganov/llama.cpp:full \ --hf-repo "THUDM/cogvlm2-llama3-chat-19B" \ --hf-file "cogvlm2-llama3-chat-19B-Q4_K_M.gguf" \ -c 4096 -ngl 99

Your hardware

More models your Radeon Pro W6800 32GB can run

ModelParamsGradeDecodeCapabilities
👁 Alibaba
Qwen3-Coder 30B A3B Instruct		30.5BS43.4 tok/s
👁 Alibaba
Qwen 3.5 27B		27BS18.8 tok/s
👁 Alibaba
Qwen 3.6 27B		27BS14.3 tok/s
👁 Alibaba
Qwen 3.6 35B A3B		35BS36.4 tok/s
👁 Alibaba
Qwen3-VL 30B A3B Instruct		30BS44.8 tok/s

Frequently asked questions

See all results for Radeon Pro W6800 32GBSee all hardware for CogVLM2 19B