BlackDolphin 12B — Hardware Requirements & GPU Compatibility
Chat
75 downloads
2 likesSpecifications
- Publisher
- Naphula
- Family
- Phi
- Parameters
- 12.2B
- Architecture
- MistralForCausalLM
- Context Length
- 1,024,000 tokens
- Vocabulary Size
- 131,072
- Release Date
- 2026-03-14
Get Started
HuggingFace
How Much VRAM Does BlackDolphin 12B Need?
Select a quantization to see compatible GPUs below.
| Quantization | Bits | VRAM | + Context | File Size | Quality |
|---|---|---|---|---|---|
| BF16 | 16.00 | 25.2 GB | 234.5 GB | 24.50 GB | Brain floating point 16 — preferred for training |
Which GPUs Can Run BlackDolphin 12B?
BF16 · 25.2 GBBlackDolphin 12B (BF16) requires 25.2 GB of VRAM to load the model weights. For comfortable inference with headroom for KV cache and system overhead, 33+ GB is recommended. Using the full 1024K context window can add up to 209.3 GB, bringing total usage to 234.5 GB. 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 BlackDolphin 12B?
BF16 · 25.2 GB15 devices with unified memory can run BlackDolphin 12B, 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 BlackDolphin 12B need?
BlackDolphin 12B requires 25.2 GB of VRAM at BF16. Full 1024K context adds up to 209.3 GB (234.5 GB total).
VRAM = Weights + KV Cache + Overhead
Weights = 12.2B × 16 bits ÷ 8 = 24.5 GB
KV Cache + Overhead ≈ 0.7 GB (at 2K context + ~0.3 GB framework)
KV Cache + Overhead ≈ 210 GB (at full 1024K context)
VRAM usage by quantization
BF1625.2 GBBF16 + full context234.5 GB- Can I run BlackDolphin 12B on a Mac?
BlackDolphin 12B requires at least 25.2 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 BlackDolphin 12B locally?
Yes — BlackDolphin 12B can run locally on consumer hardware. At BF16 quantization it needs 25.2 GB of VRAM. Popular tools include Ollama, LM Studio, and llama.cpp.
- How fast is BlackDolphin 12B?
At BF16, BlackDolphin 12B can reach ~116 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 ÷ 25.2 × 0.55 = ~116 tok/s
Estimated speed at BF16 (25.2 GB)
AMD Instinct MI300X~116 tok/sNVIDIA H100 SXM~86 tok/sAMD Instinct MI250X~72 tok/sReal-world results typically within ±20%. Speed depends on batch size, quantization kernel, and software stack.
- What's the download size of BlackDolphin 12B?
At BF16, the download is about 24.50 GB.
- Which GPUs can run BlackDolphin 12B?
1 consumer GPU can run BlackDolphin 12B at BF16 (25.2 GB). Top options include NVIDIA GeForce RTX 5090.
- Which devices can run BlackDolphin 12B?
15 devices with unified memory can run BlackDolphin 12B at BF16 (25.2 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.