NVIDIA H200
by NVIDIA · paid · Last verified 2026-04-24
The NVIDIA H200 is a Hopper-generation GPU with 141GB of HBM3e memory — nearly double the H100's bandwidth — targeting inference workloads for very large models. The additional memory enables running 70B+ parameter models on fewer GPUs, significantly reducing the cost per inference token for large-scale deployments.
https://www.nvidia.com/en-us/data-center/h200/ ↗D
D—Poor
Adoption: C+Quality: B+Freshness: ACitations: FEngagement: F
Specifications
- License
- Proprietary
- Pricing
- paid
- Capabilities
- Integrations
- Use Cases
- API Available
- No
- Tags
- nvidia, hopper, gpu, hbm3e, inference, large-models, memory
- Added
- 2026-04-24
- Completeness
- 60%
Index Score
34Adoption
50
Quality
70
Freshness
80
Citations
0
Engagement
0