Interactive Utility
MiniCPM 5 (2B) VRAM Calculator
Estimate the memory needed to run MiniCPM 5 (2B) locally. Compare quantization levels and context lengths, then check the assumptions before choosing hardware.
How much VRAM does MiniCPM 5 (2B) need? MiniCPM 5 (2B) (2.5B parameters) needs about 4.0 GB of VRAM at Q4_K_M and about 7.5 GB at FP16, with an 8k-token context. Estimate from parameter count, bits per weight, KV cache and 1.5 GB of runtime overhead — not a benchmark.
Model
Custom: billion
Smaller modeled weight footprint; check model-specific quality.
Estimated memory footprint
- Weights (Q4_K_M)
- 1.5 GB
- KV cache (8k)
- 1.0 GB
- Runtime overhead
- 1.5 GB
- Estimated system RAM to load
- 8 GB
- Estimated model file size
- 1.6 GB
Compare with your available memory
Within the modeled memory
This estimate is below the usable memory entered. Actual KV cache, quantization files and runtime allocation may change the requirement.
MiniCPM 5 (2B) memory requirements by quantisation
Estimated at an 8k-token context. MiniCPM 5 (2B) is 2.5B parameters; the arithmetic and its assumptions are set out below.
| Quantisation | Weights | Total VRAM |
|---|---|---|
| FP16 (unquantised) | 5.0 GB | 7.5 GB |
| Q8_0 (8-bit) | 2.7 GB | 5.2 GB |
| Q6_K (6-bit) | 2.1 GB | 4.6 GB |
| Q5_K_M (5-bit) | 1.8 GB | 4.3 GB |
| Q4_K_M (4-bit) | 1.5 GB | 4.0 GB |
| Q3_K_M (3-bit) | 1.2 GB | 3.7 GB |
Arithmetic over stated assumptions, not a benchmark. Weights are parameter count times bits-per-weight; the KV cache term assumes grouped-query attention rather than this model's verified KV-head count; 1.5GB is reserved for runtime overhead. Confirm the model configuration and usable device memory before purchasing hardware.
Size a related model
Embed this MiniCPM 5 (2B) Calculator
Embed a calculator prefilled with this model's parameter count in your blog, documentation, or internal wiki.
<iframe src="https://fathomlayer.com/embed/hardware-calculator?p=2.5" width="100%" height="600" frameborder="0" style="border-radius: 12px; border: 1px solid rgba(255,255,255,0.1);"></iframe>
How this math works
Local inference depends on several constraints. Memory capacity affects whether a model loads; memory bandwidth, compute and runtime settings also affect generation speed.
- Model Weights: At FP16 (unquantized), every 1 Billion parameters requires ~2GB of VRAM. At Q4_K_M (4-bit quantization), that drops to about 0.6GB per 1B parameters.
- KV Cache (Context Window): As you feed text into the model, it stores attention states in memory. A 32k context window on a 70B model requires several extra Gigabytes of RAM independent of the model weights.
- Overhead: This estimate reserves 1.5 GB for the runtime. Actual allocation depends on the device, framework and workload, so compare against usable rather than advertised memory.