NVIDIA GB10
Ranked 9 of 18 mini PCs by recommenders
Recommended for running local AI models and Linux development in a small box; the most common caveat is high price and slow memory bandwidth.
Recommended for running local AI models and Linux development in a small box; the most common caveat is high price and slow memory bandwidth.
29 people gave an opinion
- recommend it
- 16
- criticize it
- 13
- approval share
- 55%
Recommended in 8 of 13 months
- November 2025: not recommended
- December 2025: recommended
- January 2026: not recommended
- February 2026: recommended
- March 2026: not recommended
- April 2026: recommended
- May 2026: recommended
- June 2026: recommended
- July 2026: not recommended
- August 2026: recommended
- September 2026: recommended
- October 2026: recommended
- November 2026: not recommended (month in progress)
Why it doesn’t carry the badge
- 16 people recommend itNeeds 10 or more, met.
- 1.2 recommenders for every criticNeeds 2 to 1 or better, not met.
- Recommended in 8 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 local AI models well
“For personal needs I use a local Qwen3.8-Next-Flash setup on a GB10 cluster.”
Comment from 12 Sep 2026 in “Ask HN: What default model do you use and why?” “It is great for inference for single user/single session.”
Comment from 1 Jun 2026 in “Nvidia announces new AI chip for personal computers” “unreasonably effective on the GB10 IMO”
Comment from 17 Jun 2026 in “Running local models is good now” - Comment from 19 Oct 2026 in “How much intelligence can you run locally?”
Cheap enough to cluster
“new AI desktops with GB10s. They are relatively cheap and you can cluster them”
Comment from 23 Jun 2026 in “Unsloth GLM-5.2 – How to Run Locally” “For personal needs I use a local Qwen3.8-Next-Flash setup on a GB10 cluster.”
Comment from 12 Sep 2026 in “Ask HN: What default model do you use and why?” Efficient power use
“I mean the GB10 is pretty efficient for the power it has, but imho is nowhere near the power efficiency of Apple Silicon”
Comment from 1 Jun 2026 in “Nvidia announces new AI chip for personal computers” “gb10 uses way less power.”
Comment from 19 Dec 2025 in “1.5 TB of VRAM on Mac Studio – RDMA over Thunderbolt 5”
What people warn about
Each mark is one comment
Memory bandwidth is slow
“GB10 boxes have way more compute than they have memory bandwidth”
Comment from 21 Sep 2026 in “MiMo v2.6” “memory bandwidth is not just slower than on a GPU, as expected, but is significantly slower than Apple’s unified memory”
Comment from 6 Jun 2026 in “Nvidia is proposing a beast of a CPU system for Windows PCs” “the GB10 is incredibly bandwith starved. You get 128gb ram, but only 270GB/s bandwidth”
Comment from 27 Mar 2026 in “Apple discontinues the Mac Pro with no plans for future hardware” Expensive compared with alternatives
“The GB10 itself is pretty good and I love using mine for broad Linux development. But it's too expensive for consumer level pricing”
Comment from 1 Jun 2026 in “Nvidia announces new AI chip for personal computers” “The GB10 workstations are not that competitive in price with M5 MacBooks and the Mac Studio”
Comment from 16 Mar 2026 in “Analyzing Nvidia GB10's GPU” “at about $4k per GB10, that's $36k for a 1TB cluster”
Comment from 23 Jun 2026 in “Unsloth GLM-5.2 – How to Run Locally” - Comment from 23 Sep 2026 in “What Sun got wrong”
Slower than Apple Silicon
“I mean the GB10 is pretty efficient for the power it has, but imho is nowhere near the power efficiency of Apple Silicon”
Comment from 1 Jun 2026 in “Nvidia announces new AI chip for personal computers” “It is not even close. Like 0.75x SPECint 2017 of the M5.”
Comment from 1 Jun 2026 in “MacBook Pro Rival with the Nvidia Powered Surface Laptop Ultra”
Other mini PCs people recommend
The four most recommended other mini PCs, 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 |
|---|---|---|---|---|---|---|
| Apple Mac mini | 390 | 160 | 71% | 13 | 2 yrs42 owners | Approved |
| NVIDIA DGX Spark | 170 | 121 | 58% | 12 | 6 mo16 owners | — |
| AMD Strix Halo | 130 | 83 | 61% | 13 | 11 mo8 owners | — |
| Framework Desktop | 93 | 26 | 78% | 12 | 3 mo9 owners | Approved |
| NVIDIA GB10This page | 16 | 13 | 55% | 8 | — | — |
What people said
All 34 comments that mention it, newest first. Each links to the original comment.
“~50-60tk/s on GB10 128Gb VRAM”
“to have first class k8s support”
“the price through our Rep was 6k a piece no extras”
“The GB10 having so much compute is great for prefill too”
“GB10 boxes have way more compute than they have memory bandwidth”
“after setting that up, it has been pretty smooth”
“For personal needs I use a local Qwen3.8-Next-Flash setup on a GB10 cluster.”
“Power consumption though will be something like 200W idle so that's a bummer”
“Superchip anyone, with 2+ year old designs”
“I waste a lot of time waiting for answers and correcting errors”
“they'll still be slow and a little dumber than what you're used to”
“at about $4k per GB10, that's $36k for a 1TB cluster”
“new AI desktops with GB10s. They are relatively cheap and you can cluster them”
“unreasonably effective on the GB10 IMO”
“memory bandwidth is not just slower than on a GPU, as expected, but is significantly slower than Apple’s unified memory”
“It is not even close. Like 0.75x SPECint 2017 of the M5.”
“I mean the GB10 is pretty efficient for the power it has, but imho is nowhere near the power efficiency of Apple Silicon”
“It is great for inference for single user/single session.”
“The GB10 itself is pretty good and I love using mine for broad Linux development. But it's too expensive for consumer level pricing”
“I am liable to go grab a GB10 over some PC tower”
“the GB10 is better described as “Hopper with Blackwell characteristics””
“which might made gb10 look much more capable for in-house training”
“There are tons of features that are not implemented for GB10”
“the GB10 is incredibly bandwith starved. You get 128gb ram, but only 270GB/s bandwidth”
“The GB10 has been hamstrung by this a bit, too.”
“The GB10 workstations are not that competitive in price with M5 MacBooks and the Mac Studio”
“that doesn't change the reality I experience actually using the thing”
“It has very poor compute.”
“the CUDA stack is still king”
“I'm building and using this machine daily for building and using applications with LLMs, TTS, STT, ASR, and image generation.”
“unlike GB10, I'm not stuck with ARM cores and the ARM software ecosystem”
“running a bunch of Steam games flawlessly on a Aarch64 NVIDIA GB10 machine”
“gb10 uses way less power.”
“they basically asked MediaTek to build a CPU with an NV-Link I/O”
Counts come from 1,795,640 Hacker News comments, November 2025 to November 2026. How we count. Report a mistake.