Cogvlm2 Video Llama3 Chat — Hardware Requirements & GPU Compatibility
Chat
Cogvlm2 Video Llama3 Chat is a 12.5B-parameter open language model from zai-org in the Llama 3 family. It supports a context window of up to 2,048 tokens. At BF16 it needs about 27.52 GB of VRAM — see which GPUs and Macs can run it below.
133 downloads
56 likesSpecifications
- Publisher
- zai-org
- Family
- Llama 3
- Parameters
- 12.5B
- Architecture
- CogVLMVideoForCausalLM
- Context Length
- 2,048 tokens
- Vocabulary Size
- 128,256
- Release Date
- 2024-07-24
- License
- Other
Get Started
HuggingFace
How Much VRAM Does Cogvlm2 Video Llama3 Chat Need?
Select a quantization to see compatible GPUs below.
| Quantization | Bits | VRAM | + Context | File Size | Quality |
|---|---|---|---|---|---|
| BF16 | 16.00 | 27.5 GB | — | 25.02 GB | Brain floating point 16 — preferred for training |
Which GPUs Can Run Cogvlm2 Video Llama3 Chat?
BF16 · 27.5 GBCogvlm2 Video Llama3 Chat (BF16) requires 27.5 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 36+ GB is recommended. 1 GPU can run it, including NVIDIA GeForce RTX 5090.
All compatible consumer-level GPUs are running near their VRAM limit. You may also want to consider professional GPUs (e.g., NVIDIA A100, H100) which offer significantly more VRAM. For more headroom and better throughput, consider a multi-GPU configuration with tensor parallelism (supported by tools like vLLM, llama.cpp, or text-generation-inference).
Decent
— Enough VRAM, may be tightWhich Devices Can Run Cogvlm2 Video Llama3 Chat?
BF16 · 27.5 GB15 devices with unified memory can run Cogvlm2 Video Llama3 Chat, including NVIDIA DGX H100, NVIDIA DGX A100 640GB, Mac Studio M4 Max (36 GB).
Runs great
— Plenty of headroomDecent
— Enough memory, may be tightRelated Models
Frequently Asked Questions
- How much VRAM does Cogvlm2 Video Llama3 Chat need?
Cogvlm2 Video Llama3 Chat requires 27.5 GB of VRAM at BF16.
VRAM = Weights + KV Cache + Overhead
Weights = 12.5B × 16 bits ÷ 8 = 25 GB
KV Cache + Overhead ≈ 2.5 GB (at 2K context + ~0.3 GB framework)
VRAM usage by quantization
BF1627.5 GB- Can I run Cogvlm2 Video Llama3 Chat on a Mac?
Cogvlm2 Video Llama3 Chat requires at least 27.5 GB at BF16, which exceeds the unified memory of most consumer Macs. You would need a Mac Studio or Mac Pro with a high-memory configuration.
- Can I run Cogvlm2 Video Llama3 Chat locally?
Yes — Cogvlm2 Video Llama3 Chat can run locally on consumer hardware. At BF16 quantization it needs 27.5 GB of VRAM. Popular tools include Ollama, LM Studio, and llama.cpp.
- How fast is Cogvlm2 Video Llama3 Chat?
At BF16, Cogvlm2 Video Llama3 Chat can reach ~106 tok/s on AMD Instinct MI300X. Speed depends mainly on GPU memory bandwidth. Real-world results typically within ±20%.
tok/s = (bandwidth GB/s ÷ model GB) × efficiency
Example: AMD Instinct MI300X → 5300 ÷ 27.5 × 0.55 = ~106 tok/s
Estimated speed at BF16 (27.5 GB)
~106 tok/s~79 tok/s~66 tok/sReal-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
- What's the download size of Cogvlm2 Video Llama3 Chat?
At BF16, the download is about 25.02 GB.
- Which GPUs can run Cogvlm2 Video Llama3 Chat?
1 consumer GPU can run Cogvlm2 Video Llama3 Chat at BF16 (27.5 GB). Top options include NVIDIA GeForce RTX 5090.
- Which devices can run Cogvlm2 Video Llama3 Chat?
15 devices with unified memory can run Cogvlm2 Video Llama3 Chat at BF16 (27.5 GB), including Mac Mini M4 (32 GB), Mac Mini M4 Pro (48 GB), Mac Pro M2 Ultra (192 GB), Mac Studio M2 Ultra (192 GB). Apple Silicon Macs use unified memory shared between CPU and GPU, making them well-suited for local LLM inference.