Minitron 4B
NVIDIA · dense · 4B parameters · 8,192 context
Parameters
4B
Context Window
8K tokens
Architecture
Dense
Best GPU
A4000
Cheapest API
$0.06/M
Quality Score
50/100
Intelligence Brief
Minitron 4B is a 4B parameter DENSE model from NVIDIA, featuring Grouped Query Attention (GQA) with 32 layers and 3,072 hidden dimensions. With a 8,192 token context window, it supports tools, structured output, code, math, multilingual. On standardized benchmarks, it achieves MMLU 56, HumanEval 28, GSM8K 55. The most cost-effective API deployment is via nvidia-nim at $0.06/M output tokens. For self-hosted inference, A4000 delivers optimal throughput at $161/month.
Architecture Details
Memory Requirements
BF16 Weights
8.0 GB
FP8 Weights
4.0 GB
INT4 Weights
2.0 GB
GPU Compatibility Matrix
Minitron 4B is compatible with 98% of GPU configurations across 41 GPUs at 3 precision levels.
GPU Recommendations
BF16 · 1 GPU · vllm
100/100
score
Throughput
302.4 tok/s
Latency (ITL)
3.3ms
Est. TTFT
1ms
Cost/Month
$161
Cost/M Tokens
$0.20
BF16 · 1 GPU · vllm
100/100
score
Throughput
483.9 tok/s
Latency (ITL)
2.1ms
Est. TTFT
0ms
Cost/Month
$304
Cost/M Tokens
$0.24
BF16 · 1 GPU · vllm
100/100
score
Throughput
340.2 tok/s
Latency (ITL)
2.9ms
Est. TTFT
1ms
Cost/Month
$237
Cost/M Tokens
$0.27
Deployment Options
API Deployment
nvidia-nim
$0.06/M
output tokens
Single GPU
A4000
$161/mo
Min VRAM: 4 GB
Multi-GPU
RTX 3070 x2
434.9 tok/s
TP· $171/mo
API Pricing Comparison
| Provider | Input $/M | Output $/M | Badges |
|---|---|---|---|
| nvidia-nim | $0.06 | $0.06 | Cheapest |
Cost Analysis
| Provider | Input $/M | Output $/M | ~Monthly Cost |
|---|---|---|---|
| nvidia-nimBest Value | $0.06 | $0.06 | $1 |
Cost per 1,000 Requests
Short (500 tok)
$0.04
via nvidia-nim
Medium (2K tok)
$0.17
via nvidia-nim
Long (8K tok)
$0.60
via nvidia-nim
Performance Estimates
Throughput by GPU
VRAM Breakdown (A4000, BF16)
Precision Impact
bf16
8.0 GB
weights/GPU
~302.4 tok/s
fp8
4.0 GB
weights/GPU
int4
2.0 GB
weights/GPU
Quality Benchmarks
Capabilities
Features
Supported Frameworks
Supported Precisions
Where to Deploy Minitron 4B
Self-Hosted Infrastructure
Similar Models
Nemotron Mini 4B
4B params · dense
Quality: 48
from $0.06/M
Minitron 8B
8B params · dense
Quality: 62
from $0.10/M
Qwen 3 4B
4B params · dense
Quality: 57
from $0.10/M
Phi 3 Mini 3.8B
3.8B params · dense
Quality: 64
Phi 3.5 Vision
4.2B params · dense
Quality: 50
Frequently Asked Questions
How much VRAM does Minitron 4B need for inference?
Minitron 4B requires approximately 8.0 GB of VRAM at BF16 precision, 4.0 GB at FP8, or 2.0 GB at INT4 quantization. Additional VRAM is needed for KV-cache (49152 bytes per token) and activations (~0.50 GB).
What is the best GPU for Minitron 4B?
The top recommended GPU for Minitron 4B is the A4000 using BF16 precision. It achieves approximately 302.4 tokens/sec at an estimated cost of $161/month ($0.20/M tokens). Score: 100/100.
How much does Minitron 4B inference cost?
Minitron 4B API inference starts from $0.06/M input tokens and $0.06/M output tokens. Self-hosted inference costs depend on your GPU configuration — use our ROI calculator for a detailed breakdown.