OCR / VisionDeepSeekMIT

DeepSeek Janus 7B

deepseek-ai/Janus-7B

Parameters

7B

Context

4K

Downloads

0

Likes

0

Architecture

transformer

View on HuggingFace

GPU Requirements

VRAM / RAM / disk estimates by quantization and usage scenario

QuantizationMinimumRecommendedProduction
FP1616bit
20.1 GB
RAM 32.0 · Disk 59.3
23.0 GB
RAM 34.5 · Disk 85.3
39.6 GB
RAM 59.4 · Disk 124.3
FP88bit
11.8 GB
RAM 32.0 · Disk 49.2
14.4 GB
RAM 32.0 · Disk 75.2
30.6 GB
RAM 45.9 · Disk 114.2
INT88bit
11.8 GB
RAM 32.0 · Disk 49.2
14.4 GB
RAM 32.0 · Disk 75.2
30.6 GB
RAM 45.9 · Disk 114.2
AWQ4bit
7.9 GB
RAM 32.0 · Disk 44.3
10.2 GB
RAM 32.0 · Disk 70.3
26.3 GB
RAM 39.4 · Disk 109.3
GPTQ4bit
7.9 GB
RAM 32.0 · Disk 44.3
10.3 GB
RAM 32.0 · Disk 70.3
26.3 GB
RAM 39.5 · Disk 109.3
GGUF4bit
8.0 GB
RAM 32.0 · Disk 44.5
10.4 GB
RAM 32.0 · Disk 70.5
26.4 GB
RAM 39.7 · Disk 109.5

GPU Recommendations

Best ValueBEST

1× RTX 3080 Ti

12 GB VRAM

$79.5/mo

≈ 269.8 tok/s

Effective production throughput ≈ 188.9 tok/s

score 0.6677

Deploy on vast ↗
Cheapest

1× Tesla V100

16 GB VRAM

$21.1/mo

≈ 0 tok/s (N/A — benchmark unavailable)

Effective production throughput ≈ 0 tok/s

score 0.6599

Deploy on vast ↗
Performance

8× H100 SXM

640 GB VRAM

$12,249/mo

≈ 3171.6 tok/s

Effective production throughput ≈ 2220.1 tok/s

score 0.1866

Deploy on vast ↗
Min Complexity

1× RTX 3090

24 GB VRAM

$88.4/mo

≈ 276.9 tok/s

Effective production throughput ≈ 193.8 tok/s

score 0.6084

Deploy on vast ↗

💬 Community — real deployment experience

0 Articles · 0 Benchmarks

View Community →

GPUs that fit (single card)

Tesla V100

16 GB VRAM

$21.1

/mo · vast

RTX 4070S Ti

16 GB VRAM

$50.3

/mo · vast

RTX 3080 Ti

12 GB VRAM

$79.5

/mo · vast

RTX 3090

24 GB VRAM

$88.4

/mo · vast

RTX 5070

12 GB VRAM

$138

/mo · vast

AmciHub

AI Model Deployment & Compute Intelligence — analyze AI model requirements, GPU performance, cloud pricing and deployment costs to find the right deployment solution.

© 2026 AmciHub. All rights reserved. Model → Requirement → GPU → Benchmark → Cloud → Cost → Recommendation