Nvidia Nemotron 3.5 Lightning
Shares rank 31 of 39 open-weight language models by recommenders
Fast local model praised for decent code and an open training pipeline; critics say it overthinks and is weaker than comparable 27B and 35B models.
Fast local model praised for decent code and an open training pipeline; critics say it overthinks and is weaker than comparable 27B and 35B models.
13 people gave an opinion
- recommend it
- 9
- criticize it
- 4
- approval share
- 69%
Recommended in 2 of 13 months
- November 2025: not recommended
- December 2025: not recommended
- January 2026: not recommended
- February 2026: not recommended
- March 2026: not recommended
- April 2026: not recommended
- May 2026: not recommended
- June 2026: not recommended
- July 2026: not recommended
- August 2026: recommended
- September 2026: not recommended
- October 2026: recommended
- November 2026: not recommended (month in progress)
Why it doesn’t carry the badge
- 9 people recommend itNeeds 10 or more, not met.
- 2.3 recommenders for every criticNeeds 2 to 1 or better, met.
- Recommended in 2 of 13 monthsNeeds 4 or more, not 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
Fast on consumer hardware
“Nemotron 3.5 Lightning in the MXFP4 quantization absolutely flies on the Strix Halo at 65-80 t/s, but it's dumb.”
Comment from 19 Aug 2026 in “Qwen 3.8 27B is excellent, but it defaults to overthinking things” “Runs fast on my Mac - ~100 tokens/second - but it's a bit of an over-thinker.”
Comment from 11 Aug 2026 in “Nvidia Nemotron 3.5 Lightning” Generally good experiences, no major problems
“no bad experiences except for running slowly”
Comment from 11 Aug 2026 in “Nvidia Nemotron 3.5 lightning and Nemo Switchyard” “the new Nemotron model is good”
Comment from 10 Oct 2026 in “Nvidia in talks to acquire US 'open' model startup Reflection AI” Fun, good at fanfic
“its very fun (but sadly a worse model than 27B”
Comment from 11 Aug 2026 in “Nvidia Nemotron 3.5 lightning and Nemo Switchyard” “Better than average at fanfic really”
Comment from 12 Aug 2026 in “Nvidia Nemotron 3.5 lightning and Nemo Switchyard”
What people warn about
Each mark is one comment
Overthinks and second-guesses itself
“It writes pretty good code; it does seem to second-guess itself in thinking traces”
Comment from 12 Aug 2026 in “Nvidia Nemotron 3.5 Lightning” “Runs fast on my Mac - ~100 tokens/second - but it's a bit of an over-thinker.”
Comment from 11 Aug 2026 in “Nvidia Nemotron 3.5 Lightning” Weaker than 27B and 35B models
“we can train LORAs for 35B as well, why use this (according to benchmarks) worse model for customization”
Comment from 11 Aug 2026 in “Nvidia Nemotron 3.5 lightning and Nemo Switchyard” “its very fun (but sadly a worse model than 27B”
Comment from 11 Aug 2026 in “Nvidia Nemotron 3.5 lightning and Nemo Switchyard” Fails badly on some tasks
“are, well, terrible at this. They just couldn't get the job done at all.”
Comment from 12 Aug 2026 in “Nvidia Nemotron 3.5 lightning and Nemo Switchyard” “Bad, bad failure and no amount of nudging helps it see the point.”
Comment from 12 Aug 2026 in “Nvidia Nemotron 3.5 Lightning”
Other open-weight language models people recommend
The four most recommended other open-weight language models, 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 |
|---|---|---|---|---|---|---|
| Qwen 3.8 27B | 1,173 | 426 | 73% | 13 | 2 mo32 owners | Approved |
| Z.ai GLM | 1,043 | 384 | 73% | 13 | 2 mo52 owners | Approved |
| Kimi | 931 | 359 | 72% | 13 | 1 mo26 owners | Approved |
| Gemma 4 | 713 | 249 | 74% | 13 | 1 mo14 owners | Approved |
| Nvidia Nemotron 3.5 LightningThis page | 9 | 4 | 69% | 2 | — | — |
What people said
All 15 comments that mention it, newest first. Each links to the original comment.
“the new Nemotron model is good”
“Nemotron 3.5 Lightning in the MXFP4 quantization absolutely flies on the Strix Halo at 65-80 t/s, but it's dumb.”
“Better than average at fanfic really”
“Bad, bad failure and no amount of nudging helps it see the point.”
“It writes pretty good code; it does seem to second-guess itself in thinking traces”
“are, well, terrible at this. They just couldn't get the job done at all.”
“its very fun (but sadly a worse model than 27B”
“no bad experiences except for running slowly”
“we can train LORAs for 35B as well, why use this (according to benchmarks) worse model for customization”
“Runs fast on my Mac - ~100 tokens/second - but it's a bit of an over-thinker.”
“That SWEBench drop going from bf16 to fp4 is massive”
“Nvidia just throwing something "for peasants" to stay relevant.”
“this is an entirely open source training pipeline, this is quite impressive”
“Nice cadence of releases by the Nemotron team”
“nemotron models feel to me a bit less benchmaxxed / "stubborn"”
Counts come from 1,795,640 Hacker News comments, November 2025 to November 2026. How we count. Report a mistake.