NVIDIA H200
Shares rank 8 of 11 datacenter GPUs by recommenders
Datacenter GPU that people recommend for running large open-source models unquantized, mostly by renting cloud instances by the hour. Critics call it overpriced, with limited concurrency and throughput.
Datacenter GPU that people recommend for running large open-source models unquantized, mostly by renting cloud instances by the hour. Critics call it overpriced, with limited concurrency and throughput.
13 people gave an opinion
- recommend it
- 8
- criticize it
- 5
- approval share
- 62%
Recommended in 5 of 13 months
- November 2025: not recommended
- December 2025: not recommended
- January 2026: recommended
- February 2026: recommended
- March 2026: not recommended
- April 2026: not recommended
- May 2026: not recommended
- June 2026: recommended
- July 2026: not recommended
- August 2026: not recommended
- September 2026: recommended
- October 2026: recommended
- November 2026: not recommended (month in progress)
Why it doesn’t carry the badge
- 8 people recommend itNeeds 10 or more, not met.
- 1.6 recommenders for every criticNeeds 2 to 1 or better, not met.
- Recommended in 5 of 13 monthsNeeds 4 or more, met.
The same three rules apply to every product, and the badge can’t be bought. How the badge works
What people say about it
The reasons people give, with the number of comments that give each one. Open a reason to read what they said.
Why people recommend it
Each mark is one comment
Runs large open models unquantized
“to run unquantized open source models with abliteration/heretic transformations”
Comment from 5 Oct 2026 in “Anthropic reported diary entry to police, woman faces felony charge” “I ran glm 5.2 on rented 8x h200 it could only do 2x concurrency at a cost of $40 an hour”
Comment from 21 Jun 2026 in “Two Qwen3 models on one DGX Spark: the residency math” Renting by the hour is practical
“It's a depreciating asset. You rent one.”
Comment from 16 Jun 2026 in “Running local models is good now” “spin up 4xH200s at ~$9/hr on spot”
Comment from 12 Sep 2026 in “DeepSeek v4.1 Flash Uncensored” Faster and cheaper than H100 setups
“8xH200s will be a lot cheaper than 16xH100s, and faster too.”
Comment from 27 Jan 2026 in “Kimi Released Kimi K2.5, Open-Source Visual SOTA-Agentic Model” “H200 is a major step up over H100.”
Comment from 11 Feb 2026 on Hacker News
What people warn about
Each mark is one comment
Overpriced for what it delivers
“H200's are not that cost effective and saturate too fast while having poorer efficiency”
Comment from 26 Jun 2026 in “Why current LLM costs are not sustainable” “The datacenter GPUs sell at an 80% margin. They're incredibly overpriced.”
Comment from 24 May 2026 in “Memory has grown to nearly two-thirds of AI chip component costs” Limited concurrency and throughput
“I ran glm 5.2 on rented 8x h200 it could only do 2x concurrency at a cost of $40 an hour”
Comment from 21 Jun 2026 in “Two Qwen3 models on one DGX Spark: the residency math” “performance/throughput will be roughly halved (closer to 2/3) and free context will be more limited with 8xH200”
Comment from 27 Jan 2026 in “Kimi Released Kimi K2.5, Open-Source Visual SOTA-Agentic Model” “H200's are not that cost effective and saturate too fast while having poorer efficiency”
Comment from 26 Jun 2026 in “Why current LLM costs are not sustainable” Needs a rack and is expensive
“That GPU costs 25k which means you really should have a rack to put it in. It's not realistic.”
Comment from 16 Jun 2026 in “Running local models is good now” “The datacenter GPUs sell at an 80% margin. They're incredibly overpriced.”
Comment from 24 May 2026 in “Memory has grown to nearly two-thirds of AI chip component costs”
Other datacenter GPUs people recommend
The four most recommended other datacenter GPUs, with the same counts for each, so you can compare like for like. The tick on each bar marks the 2 to 1 line the badge needs.
| Product | Recommend | Criticize | Approval | Months | Owned | Badge |
|---|---|---|---|---|---|---|
| NVIDIA H100 | 20 | 16 | 56% | 11 | — | — |
| NVIDIA Tesla V100 | 16 | 4 | 80% | 8 | — | Approved |
| AMD Instinct MI355X | 14 | 5 | 74% | 7 | — | Approved |
| Nvidia A100 | 10 | 4 | 71% | 8 | — | Approved |
| NVIDIA H200This page | 8 | 5 | 62% | 5 | — | — |
What people said
11 of the 14 comments that mention it, newest first. Each links to the original comment.
“to run unquantized open source models with abliteration/heretic transformations”
“you can absolutely buy 16 x H200SXMs, with the appropriate boards to take them”
“spin up 4xH200s at ~$9/hr on spot”
“H200's are not that cost effective and saturate too fast while having poorer efficiency”
“I ran glm 5.2 on rented 8x h200 it could only do 2x concurrency at a cost of $40 an hour”
“It's a depreciating asset. You rent one.”
“That GPU costs 25k which means you really should have a rack to put it in. It's not realistic.”
“The datacenter GPUs sell at an 80% margin. They're incredibly overpriced.”
“H200 is a major step up over H100.”
“performance/throughput will be roughly halved (closer to 2/3) and free context will be more limited with 8xH200”
“8xH200s will be a lot cheaper than 16xH100s, and faster too.”
Counts come from 1,795,640 Hacker News comments, November 2025 to November 2026. How we count. Report a mistake.