CogVLM2 19B needs ~13.9 GB VRAM. Intel Arc B580 12GB has 12.0 GB. With Q3_K_S quantization, expect ~13 tok/s.
Operating mode
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.
Select quantization to explore
4.1 GB over capacity — needs offload or smaller quantization
Fit status
Too heavy
Decode
8.5 tok/s
TTFT
22803 ms
Safe context
4K
Memory
16.1 GB / 12.0 GB
Offload
30%
It fits through host-memory offload, and offload is the main reason performance drops.
CPU or host-memory offload is active
About 10% of the working set spills out of accelerator memory, which usually hurts latency and sustained decode throughput.
Very little memory headroom
You can run the model, but there is not much room left for longer context, bigger batches, extra apps, or future model updates.
Runtime ecosystem is narrower than CUDA
Intel GPUs can look attractive on memory per dollar, but local AI tooling, kernels, and model coverage are still broader and easier on CUDA today.
Remove offload with more accelerator memory
Prioritize a GPU or unified-memory tier that fits the whole model natively. Removing offload usually helps more than small compute gains.
Prefer CUDA if you want the path of least resistance
If your goal is maximum runtime coverage, easier troubleshooting, and better support for new local AI releases, CUDA is usually still the safer upgrade path.
Buy headroom, not only minimum fit
A slightly larger memory tier gives you safer context growth and makes the recommendation more future-proof.
| Workload | Grade | Fit | Decode | TTFT | Context |
|---|---|---|---|---|---|
| Chat | F | Too heavy | 10.0 tok/s | 10594 ms | 4K |
| Coding | F | Too heavy | 8.5 tok/s | 22803 ms | 4K |
| Agentic Coding | F | Too heavy | 6.4 tok/s | 44217 ms | 4K |
| Reasoning | F | Too heavy | 8.5 tok/s | 26950 ms | 4K |
| RAG | F | Too heavy | 6.4 tok/s | 55271 ms | 4K |
How CogVLM2 19B (19B params) fits at each quantization level on Intel Arc B580 12GB (12.0 GB usable).
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_KBest for your GPU | 2 | 7.4 GB | Low | A85 |
Q3_K_S | 3 | 9.3 GB | Low | F0 |
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 99Upgrade options
Intel Arc A770 16GBBudget pickMakes the model fit on the accelerator instead of staying completely out of reach.
Removes host-memory offload, which is usually the single biggest latency and throughput win.
~$349 MSRP
Intel Arc Pro B50 16GBBest valueMakes the model fit on the accelerator instead of staying completely out of reach.
Removes host-memory offload, which is usually the single biggest latency and throughput win.
~$399 MSRP
Intel Arc Pro B60 24GBIntel upgradeMakes the model fit on the accelerator instead of staying completely out of reach.
Removes host-memory offload, which is usually the single biggest latency and throughput win.
~$599 MSRP
Yes, Intel Arc B580 12GB can run CogVLM2 19B at Q3_K_S quantization (Very compromised (needs ~1.2 GB host RAM)). The recommended Q4_K_M requires 16.1 GB which exceeds available memory, but at Q3_K_S it needs only 13.9 GB. Expected decode speed: 13.4 tok/s.
CogVLM2 19B (19B parameters) requires approximately 16.1 GB at Q4_K_M quantization. On Intel Arc B580 12GB, it fits at Q3_K_S using 13.9 GB.
The recommended quantization is Q4_K_M, but on Intel Arc B580 12GB the best fitting quantization is Q3_K_S, which uses 13.9 GB.
On Intel Arc B580 12GB, CogVLM2 19B achieves approximately 13.4 tokens per second decode speed with a time-to-first-token of 14434ms using Q3_K_S quantization.
For coding workloads, CogVLM2 19B on Intel Arc B580 12GB receives a F grade with 8.5 tok/s and 4K context.
On Intel Arc B580 12GB, CogVLM2 19B can safely use up to 4K tokens of context at Q3_K_S quantization. The model's official context limit is 8K, but available memory constrains the safe maximum.
Paste this snippet into any page to show a live fit card.
<iframe src="https://willitrunai.com/embed/cogvlm2-19b-on-arc-b580-12gb" width="400" height="180" frameborder="0" style="border:none;border-radius:12px;overflow:hidden;" title="Will It Run AI — fit result"></iframe>
Preview:
| 4 |
10.6 GB |
| Medium |
| F0 |
Q4_K_M | 4 | 11.6 GB | Medium | F0 |
Q5_K_M | 5 | 13.7 GB | High | F0 |
Q6_K | 6 | 15.6 GB | High | F0 |
Q8_0 | 8 | 20.3 GB | Very High | F0 |
F16 | 16 | 38.9 GB | Maximum | F0 |
Remove offload with more accelerator memory. Prioritize a GPU or unified-memory tier that fits the whole model natively. Removing offload usually helps more than small compute gains.
Often yes, if your goal is the easiest setup and the widest runtime support. Intel can offer attractive memory capacity, but CUDA still tends to win on tooling maturity, guides, kernels, and model coverage for local AI.