BERT
Hacker ApprovedShares rank 36 of 72 language models by recommenders
Fast, cheap model that people fine-tune for classification and sentiment analysis, often competitive with larger LLMs. Caveat: it needs good data and expertise.
Fast, cheap model that people fine-tune for classification and sentiment analysis, often competitive with larger LLMs. Caveat: it needs good data and expertise.
23 people gave an opinion
- recommend it
- 19
- criticize it
- 4
- approval share
- 83%
Recommended in 10 of 13 months
- November 2025: recommended
- December 2025: recommended
- January 2026: recommended
- February 2026: not recommended
- March 2026: recommended
- April 2026: not recommended
- May 2026: recommended
- June 2026: recommended
- July 2026: not recommended
- August 2026: recommended
- September 2026: recommended
- October 2026: recommended
- November 2026: recommended (month in progress)
Why it carries the badge
- 19 people recommend itNeeds 10 or more, met.
- 4.8 recommenders for every criticNeeds 2 to 1 or better, met.
- Recommended in 10 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
Good results when fine-tuned
“they do a fairly decent job as long as you have good data and know what you're doing”
Comment from 14 Nov 2025 in “Honda: 2 years of ml vs 1 month of prompting - heres what we learned” “you could just fine-tune BERT and get a faster model with better performance”
Comment from 22 Jun 2026 in “Good results fine tuning a local LLM like Qwen 3:0.6B to categorize questions” “Which in my experience is indeed quite effective (and easy/fast).”
Comment from 14 Nov 2025 in “Honda: 2 years of ml vs 1 month of prompting - heres what we learned” - Comment from 29 Sep 2026 in “Jeff – Jev-compatible 0.8B decision models, trained at home, ~30 ms”
- Comment from 6 Aug 2026 in “Prime Agent: A self-improving RLM agent”
- Comment from 22 Oct 2026 in “Using LLM for confidence scoring is a bad idea”
Small models beat large LLMs
“its way more useful to train small purposeful models than to use an entire frontier LLM”
Comment from 1 Nov 2026 in “Anthropic blocks users for anything touching training larger models” “A small transformer like BERT or variants is a better fit.”
Comment from 22 Jun 2026 in “Good results fine tuning a local LLM like Qwen 3:0.6B to categorize questions” “We had good small language models for decades.”
Comment from 1 Mar 2026 in “Microgpt” Fast and cheap to run
“Something like BERT could be and has been tuned to provide similar "decision making" at a similar latency and cost advantage”
Comment from 18 Sep 2026 in “OpenJev” “you could just fine-tune BERT and get a faster model with better performance”
Comment from 22 Jun 2026 in “Good results fine tuning a local LLM like Qwen 3:0.6B to categorize questions” “perform quite well in embedded applications”
Comment from 27 Sep 2026 in “I was impressed by Jev, please explain why I shouldn't be” Strong at classification and sentiment
“BERT was highly usable for classification and sentiment analysis a whopping 9 years ago”
Comment from 19 Sep 2026 in “I built non-autoregressive decision models with RL a year ago” “already known as SotA for text classification and similarity back in 2023”
Comment from 14 Nov 2025 in “Honda: 2 years of ml vs 1 month of prompting - heres what we learned” “lightweight encoder models like BERT are great at sentiment analysis”
Comment from 3 Nov 2025 in “Tongyi DeepResearch – open-source 30B MoE Model that rivals OpenAI DeepResearch”
What people warn about
Each mark is one comment
Too weak for real use
“it was only good enough to be worth investigating further, not to actually use”
Comment from 31 Mar 2026 in “Mr. Chatterbox is a Victorian-era ethically trained model” “You couldn't really get it to do useful stuff”
Comment from 19 Sep 2026 in “I built non-autoregressive decision models with RL a year ago”
Other language models people recommend
The four most recommended other 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 |
|---|---|---|---|---|---|---|
| Claude Opus | 3,331 | 2,746 | 55% | 13 | 4 mo156 owners | — |
| OpenAI GPT | 1,205 | 987 | 55% | 13 | 6 mo47 owners | — |
| Claude Sonnet | 854 | 586 | 59% | 13 | 6 mo46 owners | — |
| DeepSeek V4 Flash | 828 | 223 | 79% | 8 | 1 mo24 owners | Approved |
| BERTThis page | 19 | 4 | 83% | 10 | — | Approved |
What people said
23 of the 26 comments that mention it, newest first. Each links to the original comment.
“its way more useful to train small purposeful models than to use an entire frontier LLM”
“I've had very good luck with”
“even the largest of those models has the world knowledge of a fart to draw priors from”
“if you're finetuning, a bert base is not bad”
“we used various bert variants to satisfy our usecases and still in uses”
“perform quite well in embedded applications”
“doing a BERT run takes about 2 hours from having an empty folder”
“BERT was highly usable for classification and sentiment analysis a whopping 9 years ago”
“You couldn't really get it to do useful stuff”
“Something like BERT could be and has been tuned to provide similar "decision making" at a similar latency and cost advantage”
“used bert model with decent results”
“you could just fine-tune BERT and get a faster model with better performance”
“A small transformer like BERT or variants is a better fit.”
“seeing BERT just being able to straight up answer questions about a document”
“in NLP circles it was a ‘magical’ improvement”
“it was only good enough to be worth investigating further, not to actually use”
“We had good small language models for decades.”
“The BERT models are easy to probability calibrate too!”
“Whatever you do don't screw around with fine-tuned BERT.”
“Which in my experience is indeed quite effective (and easy/fast).”
“they do a fairly decent job as long as you have good data and know what you're doing”
“already known as SotA for text classification and similarity back in 2023”
“lightweight encoder models like BERT are great at sentiment analysis”
Counts come from 1,795,640 Hacker News comments, November 2025 to November 2026. How we count. Report a mistake.