AIXX Pedia The encyclopedia of compute hardware

AIXX Pedia › Model Comparison › H100 × H200

H100 vs H200

Model Comparison · Verdict First · Specs Side by Side · AIXX Index Side by Side · AIXX Pedia

Verdict First VERDICT

HOW TO CHOOSE · One-Paragraph Verdict
Bottleneck is budget, sub-70B-class inference, or HPC/FP64 , choose the H100 (80GB HBM3, largest ecosystem, deepest installed base); bottleneck is memory capacity (100B+ parameter models, long context, full R1), choose H200 (141GB HBM3e, RMB 5.85–7M in mainland market). Compute parity, same-generation interconnect (NVLink 4 · 900GB/s) — the entire difference is in memory.

Specs Side by Side SPECIFICATIONS

ParameterNVIDIA H100 · Hopper GenerationNVIDIA H200 · Hopper Generation
ArchitectureHopperHopper (memory-enhanced)
ProcessTSMC 4NTSMC 4N
Board Memory80GB HBM3 (640GB across 8 cards)141GB HBM3e
Memory Bandwidth3.35TB/s per card4.8TB/s per card
GPU-to-GPU InterconnectNVLink 4 · 900GB/sNVLink 4 · 900GB/s
Board Power (TBP/TGP)Up to 700W (configurable)Up to 700W (configurable)
Form Factor & CoolingHGX H100 8-card baseboard (mostly air-cooled towers · OEM liquid-cooled variants available)HGX H200 8-GPU baseboard (mostly air-cooled tower · OEM liquid-cooled variants available)
Launch / DeliveryAnnounced 2022-03 · shipping since 2022-10Announced 2023-11 · Shipping since 2024-06
China-market TimelineSubject to China export controls since 2022-10 · not on the approved list for ChinaApproved for the China-market Variant 2025-12 · Volume delivery from 2026-07
Mainland China Street Price BandApprox. 5.02M per system (2026-10-07 live market floor · 1 unit · whole-lot asking basis)¥5.85–7 million per system (2026-10-07 live marketplace · 3 units · whole-lot asking price)

Specs are based on NVIDIA official public documentation and the two entry pages; price bands reflect actual on-sale listings in this library's marketplace and fluctuate with inventory—see the entry page "price band" for the latest. A dash in both columns means the entry does not yet include this parameter.

AIXX Index Side by Side WANAN INDEX

AIXX Index (0–100): estimated by this library using a weighted public-spec model—memory 40% + interconnect 30% + ecosystem 30%, weights adjusted by scenario. 90+ flagship · 75+ Excellent · 60+ usable · below 60 not applicable. The method is transparent and verifiable and does not constitute a performance promise; figures match the published figures on the two entry pages, so this page does not score separately.

Scenario IndexH100H200
LLM Inference8588
Model Training / Fine-tuning8885
HPC / FP64 Computing92—Not rated
Cost-sensitive Inference9090
Large-memory scenarios (long context / full R1)58—Not rated
Long-context / RAG—Not rated92
Video Generation—Not rated78
Condition / Reliability / Value Retention (Experience Index)Pending hands-on testingPending hands-on testing

"—" = the machine's entry page has no rating for this scenario; this page does not fill it in. The Experience Index (condition/reliability/value retention) is pending hands-on testing—inspection batch data is being accumulated and will light up unit by unit as Inspection Reports are published.

Which to Choose for Which Scenario WHICH ONE

Choose the H100 when —
Tight budget, ecosystem first: sub-70B-class inference/fine-tuning and HPC/FP64 workloads, the most mature installed and rental market, the volume workhorse in stock trading.
Choose H200 when —
Bottleneck is memory: 141GB per card / 1.128TB per 8-card node — full R1 at FP8 (roughly 690GB) deploys on a single node, with ample headroom for long context.
Choose neither, when—
For FP4 compute and next-gen interconnect, go with the B300 (288GB HBM3e / NVLink 5); if neither fits, check the spec sheet and market price band on the entry page before deciding.

Keep Verifying LINKS

After this comparison, return to the entry pages to verify details and the latest price bands:

Footnote: This page is for reference comparison only and does not constitute a trading offer; actual price subject to the seller's quote. Data source: AIXX Compute Exchange (entry figures 2026-10; index method identical to the two entry pages, transparent and verifiable).
This Library
Editorial PolicyAbout This Library (Disclosure of Interests)Corrections Channel
Editorial Constitution Summary
Every claim sourcedDrawbacks always stated · corrections publicly loggedNo paid entries acceptedZero traffic-funneling or sales elements within entriesMarket data attached only as data modules
AIXX Family
AIXX PediaAIXX ExchangeAIXX News
Market data source: AIXX Compute Exchange · © 2026 AIXX Pedia · An AIXX Production