Whisper Small
OpenAI · dense · 0.244B parameters · 448 context
Parameters
0.244B
Context Window
0K tokens
Architecture
Dense
Best GPU
B200 SXM
Intelligence Brief
Whisper Small is a 0.244B parameter DENSE model from OpenAI, featuring Multi-Head Attention (MHA) with 12 layers and 768 hidden dimensions. With a 448 token context window, it supports multilingual. For self-hosted inference, B200 SXM delivers optimal throughput at $4261/month.
Architecture Details
Memory Requirements
BF16 Weights
0.5 GB
FP8 Weights
0.2 GB
INT4 Weights
0.1 GB
GPU Compatibility Matrix
Whisper Small is compatible with 100% of GPU configurations across 41 GPUs at 3 precision levels.
GPU Recommendations
FP8 · 1 GPU · tensorrt-llm
83/100
score
Throughput
3.5K tok/s
Latency (ITL)
0.3ms
Est. TTFT
0ms
Cost/Month
$4261
Cost/M Tokens
$0.46
FP8 · 1 GPU · tensorrt-llm
83/100
score
Throughput
3.5K tok/s
Latency (ITL)
0.3ms
Est. TTFT
0ms
Cost/Month
$4271
Cost/M Tokens
$0.46
FP8 · 1 GPU · tensorrt-llm
83/100
score
Throughput
3.5K tok/s
Latency (ITL)
0.3ms
Est. TTFT
0ms
Cost/Month
$6169
Cost/M Tokens
$0.67
Deployment Options
API Deployment
No API pricing available
Single GPU
B200 SXM
$4261/mo
Min VRAM: 0 GB
Multi-GPU
B200 SXM
3.5K tok/s
Best available config
API Pricing Comparison
No API pricing data available for this model.
Performance Estimates
Throughput by GPU
VRAM Breakdown (B200 SXM, FP8)
Precision Impact
bf16
0.5 GB
weights/GPU
fp8
0.2 GB
weights/GPU
~3.5K tok/s
Capabilities
Features
Supported Frameworks
Supported Precisions
Where to Deploy Whisper Small
Self-Hosted Infrastructure
Similar Models
Whisper Base
0.074B params · dense
Quality: 50
NV EmbedQA E5 v5
0.33B params · dense
Quality: 50
from $0.01/M
NV Retriever v1
0.33B params · dense
Quality: 50
from $0.01/M
BGE Large EN v1.5
0.335B params · dense
Quality: 50
from $0.01/M
Nomic Embed Text v1.5
0.137B params · dense
Quality: 50
from $0.01/M
Frequently Asked Questions
How much VRAM does Whisper Small need for inference?
Whisper Small requires approximately 0.5 GB of VRAM at BF16 precision, 0.2 GB at FP8, or 0.1 GB at INT4 quantization. Additional VRAM is needed for KV-cache (36864 bytes per token) and activations (~0.10 GB).
What is the best GPU for Whisper Small?
The top recommended GPU for Whisper Small is the B200 SXM using FP8 precision. It achieves approximately 3.5K tokens/sec at an estimated cost of $4261/month ($0.46/M tokens). Score: 83/100.
How much does Whisper Small inference cost?
Whisper Small inference costs vary by provider and GPU setup. Use our calculator for detailed cost estimates across all providers.