RUNLOCALAIv38
->Will it run?Best GPUCompareTroubleshootStartLearnPulseModelsHardwareToolsBench
Run check
  1. >
  2. Home
  3. /Hardware
  4. /NVIDIA H20 (96GB)
UNIT · NVIDIA · GPU
96 GB VRAMworkstation·Reviewed June 2026

NVIDIA H20 (96GB)

NVDA · HARDWARE
NVIDIA H20 (96GB)

No editorial image yet — generic vendor mark shown. Credentials in spec table below.

The China-market Hopper SKU tuned for inference: 96GB HBM3 (more than the standard H100's 80GB), 4.0 TB/s, 400W, with ~41% fewer cores than a full H100. Export-compliant and highly relevant where H100/H200 are restricted.

Released 2024·4000 GB/s memory bandwidth
RUNLOCALAI SCORE
See full leaderboard →
697/ 1000
BB-tier
Estimated
Throughput
500/ 500
VRAM-fit
200/ 200
Ecosystem
200/ 200
Efficiency
96/ 100

Sub-scores sum to 996 / 1000. Headline = 996 × 0.70 (Estimated-confidence discount) = 697. This is an algorithmic performance-tier score — distinct from, and often lower than, the editorial “Our verdict” below, which weighs value and real-world fit (especially for hardware we haven’t measured yet). How scoring works →

Extrapolated from 4000 GB/s bandwidth — 480.0 tok/s estimated. No measured benchmarks yet.

WORKLOAD FIT
Try other hardware →

Plain-English: Runs 70B comfortably — snappy enough for a coding agent; vision models supported.

7B chat✓
Comfortable
14B chat✓
Comfortable
32B chat✓
Comfortable
70B chat✓
Comfortable
Coding agent✓
Comfortable
Vision (≤8B VLM)✓
Comfortable
Long context (32K)✓
Comfortable
✓Comfortable — fits with headroom
~Tight — works, no slack
△Marginal — needs aggressive quant
✗Doesn't fit usefully

Verdicts extrapolated from catalog VRAM + bandwidth + ecosystem flags. Hover any chip for the rationale. Want measured numbers? Submit your own run with runlocalai-bench --submit.

BLK · VERDICT

Our verdict

OP · Fredoline Eruo|VERIFIED JUN 18, 2026
7.4/10

What it is

The H20 is NVIDIA's export-compliant Hopper part for the Chinese market — deliberately tuned for inference rather than training. It pairs cut-down compute (~41% fewer cores than a full H100) with an unusually large 96GB of HBM3 at 4.0 TB/s, which actually exceeds the standard H100's 80GB. At 400W it runs 30B FP16 or 70B quantized comfortably.

Relevance to local AI

For local/on-prem AI buyers in China — where H100/H200 are restricted — the H20 is often the most capable CUDA card legally available, and its 96GB makes it a genuinely strong inference GPU despite the compute cuts (inference is more memory- than compute-bound). The high VRAM-to-compute ratio is well-matched to serving, less so to training. Outside China it's largely irrelevant given access to full Hopper/Blackwell parts.

Bottom line

A niche-but-real entry: the export-compliant 96GB Hopper inference card that matters specifically to China-market local-AI deployments. Included for completeness; not relevant to buyers with access to standard H100/H200.

BLK · OVERVIEW

Overview

The China-market Hopper SKU tuned for inference: 96GB HBM3 (more than the standard H100's 80GB), 4.0 TB/s, 400W, with ~41% fewer cores than a full H100. Export-compliant and highly relevant where H100/H200 are restricted.

Retailers we'd check:Amazon

Search-fallback link — editorial hasn't yet curated a retailer URL for this card.

Some links above are affiliate links. We may earn a commission at no extra cost to you. How we make money.

BLK · SPECS

Specs

VRAM96 GB
Power draw (peak)400 W
Released2024
Backends
CUDA

Models that fit

Open-weight models small enough to run on NVIDIA H20 (96GB) with usable context.

all-MiniLM-L6-v2
0.022B · other
FLUX.1 [dev]
12B · other
Qwen 3 0.6B
0.6B · qwen
BGE Large EN v1.5
0.335B · other
Nomic Embed Text v1.5
0.137B · other
Kokoro 82M
0.082B · other
Llama 3.1 8B Instruct
8B · llama
Qwen 3 30B-A3B
30B · qwen

Frequently asked

What models can NVIDIA H20 (96GB) run?

With 96GB VRAM, the NVIDIA H20 (96GB) runs 70B models in 4-bit quantization, plus everything smaller. See the model list below for tested combinations.

Does NVIDIA H20 (96GB) support CUDA?

Yes — NVIDIA H20 (96GB) is an NVIDIA card with full CUDA support, the most mature local-AI backend. llama.cpp, Ollama, vLLM, and ExLlamaV2 all run natively.

Where next?

Buyer guides
  • Best GPU for local AI →
  • Best laptop for local AI →
  • Best Mac for local AI →
  • Best used GPU for local AI →
Troubleshooting
  • CUDA out of memory →
  • Ollama running slowly →
  • ROCm not detected →
  • Model keeps crashing →

Reviewed by RunLocalAI Editorial. See our editorial policy for how we research and verify hardware specifications.

RUNLOCALAI

Independently operated catalog for local-AI hardware and software. Hand-written verdicts. Source-cited claims. Reproducible commands when we have them.

OP·Fredoline Eruo
DIR
  • Models
  • Hardware
  • Tools
  • Benchmarks
TOOLS
  • Will it run?
  • Compare hardware
  • Cost vs cloud
  • Choose my GPU
  • Prompting kits
  • Quick answers
REF
  • All buyer guides
  • Learn local AI
  • Methodology
  • Glossary
  • Errors KB
  • Trust
EDITOR
  • About
  • Author
  • How we make money
  • Editorial policy
  • Contact
LEGAL
  • Privacy
  • Terms
  • Sitemap
MAIL · MONTHLY DIGEST
Get monthly local AI changes
Monthly recap. No spam.
DISCLOSURE

Some links on this site are affiliate links (Amazon Associates and other first-class retailers). When you buy through them, we earn a small commission at no extra cost to you. Affiliate links do not influence our verdicts — there are cards we rate highly that we don't have affiliate relationships with, and cards that sell well that we refuse to recommend. Read more →

© 2026 runlocalai.coIndependently operated
RUNLOCALAI · v38
Compare alternatives

Hardware worth comparing

The closest alternatives by price, memory bandwidth, and form factor, plus a step up and down — so you can frame the buying decision against real options.

Closest matches
Similar price, bandwidth & form factor
  • Intel Gaudi 3
    intel · 128 GB VRAM
    8.2/10
  • Intel Gaudi 2
    intel · 96 GB VRAM
    7.9/10
  • AMD Instinct MI300A (APU)
    amd · 128 GB VRAM
    10.0/10
  • AMD Instinct MI250X
    amd · 128 GB VRAM
    9.7/10
  • AMD Instinct MI300X
    amd · 192 GB VRAM
    10.0/10
  • NVIDIA H100 SXM
    nvidia · 80 GB VRAM
    10.0/10
Step up
More capable — more memory or a higher tier
  • Intel Gaudi 3
    intel · 128 GB VRAM
    8.2/10
  • AMD Instinct MI300A (APU)
    amd · 128 GB VRAM
    10.0/10
  • NVIDIA H100 NVL
    nvidia · 188 GB VRAM
    10.0/10
Step down
Lighter — cheaper or more constrained
  • AMD Instinct MI210
    amd · 64 GB VRAM
    9.8/10
  • Intel Arc Pro B60 24GB
    intel · 24 GB VRAM
    7.6/10
  • AMD Radeon RX 7900 XTX
    amd · 24 GB VRAM
    7.8/10