Tencent Hy3
Hacker ApprovedRanked 22 of 39 open-weight language models by recommenders
Recommended as a cheap, capable open-weights model that punches above its size for coding and writing. Caveats: slow responses, rate limits and errors on OpenRouter.
Recommended as a cheap, capable open-weights model that punches above its size for coding and writing. Caveats: slow responses, rate limits and errors on OpenRouter.
36 people gave an opinion
- recommend it
- 25
- criticize it
- 11
- approval share
- 69%
Recommended in 5 of 13 months
- November 2025: not recommended
- December 2025: not recommended
- January 2026: not recommended
- February 2026: not recommended
- March 2026: recommended
- April 2026: not recommended
- May 2026: not recommended
- June 2026: recommended
- July 2026: recommended
- August 2026: recommended
- September 2026: not recommended
- October 2026: recommended
- November 2026: not recommended (month in progress)
Why it carries the badge
- 25 people recommend itNeeds 10 or more, met.
- 2.3 recommenders for every criticNeeds 2 to 1 or better, 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
Cheap to use for its capability
“except for Hy3 , DeepSeek v4 Pro , and MiMo v2.5 Pro (whose API rates are cheap and/or discounted already)”
Comment from 22 Jul 2026 in “Kimi K3 Is Competitive with Fable; Kimi K3 and Fable Is SoTA” “it just works for me for most things I tried and it's exceedingly cheap”
Comment from 9 Jul 2026 in “Hy3” “genuinely good enough as daily drivers, given the costs”
Comment from 16 Jul 2026 in “Kimi K3 is now live” - Comment from 9 Jul 2026 in “Hy3”
Strong and surprisingly good overall
“Hy3 was a very strong and underrated model.”
Comment from 30 Aug 2026 in “Hy4 preview” “was surprised with how good it was”
Comment from 29 Aug 2026 in “Hy4 preview” “Hy3 have all been amazing models.”
Comment from 17 Jun 2026 in “GLM-5.2 is the new leading open weights model on Artificial Analysis” - Comment from 16 Jul 2026 in “Inkling: Our Open-Weights Model”
Strong coding and benchmark results
“Hy3's coding benchmarks hold their own against DeepSeek v4 Pro & MiMo v2.5 Pro”
Comment from 9 Jul 2026 in “Hy3” “I was also surprised to see how (relatively) good it is on coding tasks too.”
Comment from 30 Aug 2026 in “Hy4 preview” “the benchmarks are on par with Pro (or above in some cases)”
Comment from 9 Jul 2026 in “Hy3” Small for how capable it is
“Which is impressive as it's smaller and cheaper than the others.”
Comment from 10 Jul 2026 in “Hy3” “This model is shockingly small for how capable it is.”
Comment from 9 Jul 2026 in “Hy3”
What people warn about
Each mark is one comment
Slow and error-prone on OpenRouter
“My one concern over Hy3 was speed.”
Comment from 29 Aug 2026 in “Hy4 preview” “It is really slow on openrouter and I ran into many http errors.”
Comment from 9 Jul 2026 in “Hy3” Rate limits on the free tier
“free with generous limits and high availability, until it wasn't”
Comment from 10 Jul 2026 in “Hy3” “I was getting rate limited like crazy”
Comment from 9 Jul 2026 in “Hy3” Trails rival models in some tests
“The Hy3 preview has been a mediocre performer in my benchmarks of security auditing with models”
Comment from 9 Jul 2026 in “Hy3” “I don't see a reason where you would use this model over competitors.”
Comment from 9 Jul 2026 in “Hy3” “Worse than GLM-5.2, much more expensive than deepseek v4.”
Comment from 10 Jul 2026 in “Hy3”
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 |
| Tencent Hy3This page | 25 | 11 | 69% | 5 | — | Approved |
What people said
34 of the 40 comments that mention it, newest first. Each links to the original comment.
“Hy3 was a very strong and underrated model.”
“I was also surprised to see how (relatively) good it is on coding tasks too.”
“was surprised with how good it was”
“an unprecedented jump between preview and final release on Hy3, from unusable to competitive”
“My one concern over Hy3 was speed.”
“I've had some pretty decent results with Hy3 and Nemotron too, but they're not the same”
“always available, and I can't be happier”
“except for Hy3 , DeepSeek v4 Pro , and MiMo v2.5 Pro (whose API rates are cheap and/or discounted already)”
“genuinely good enough as daily drivers, given the costs”
“I tried Hy3 today and liked it.”
“Which is impressive as it's smaller and cheaper than the others.”
“Worse than GLM-5.2, much more expensive than deepseek v4.”
“free with generous limits and high availability, until it wasn't”
“it seems to stay on track much more reliably than DS4 Flash”
“hy3 makes idiotic decisions all the time”
“It is really slow on openrouter and I ran into many http errors.”
“Writes pretty engaging prose, finetunes well, now MIT licensed”
“the benchmarks are on par with Pro (or above in some cases)”
“it's just some confusingly-named LLM. Shame.”
“The Hy3 preview has been a mediocre performer in my benchmarks of security auditing with models”
“it's either benchmaxxed to hell and back or skill issue on my part”
“I was getting rate limited like crazy”
“That UI demo page is… really quite janky.”
“I don't see a reason where you would use this model over competitors.”
“Hy3's coding benchmarks hold their own against DeepSeek v4 Pro & MiMo v2.5 Pro”
“Hy3 lacks the DSv4 architecture's KV Cache efficiency.”
“It's a very good model for this size and price.”
“its still quite big unless by "local" you mean people that spend many thousands on rigs”
“This model is shockingly small for how capable it is.”
“it just works for me for most things I tried and it's exceedingly cheap”
“there's many good Chinese models and labs, glm, qwen, deepseek, kimik, mimo, longcat, minimax, hy3”
“Hy3 have all been amazing models.”
“Openweight models turned a corner around kimi 2.6, deepseek v4 pro/flash, hy3 and mimo 2.5 pro.”
“hy3 is the only way I can bear using hyprland”
Counts come from 1,795,640 Hacker News comments, November 2025 to November 2026. How we count. Report a mistake.