SubQ
Shares rank 44 of 72 language models by recommenders
People recommend SubQ for claimed long-context reasoning that is faster and cheaper than transformer models. The most common caveat is the missing technical report and no independent verification.
People recommend SubQ for claimed long-context reasoning that is faster and cheaper than transformer models. The most common caveat is the missing technical report and no independent verification.
25 people gave an opinion
- recommend it
- 11
- criticize it
- 14
- approval share
- 44%
Recommended in 3 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: recommended
- June 2026: recommended
- July 2026: not recommended
- August 2026: not recommended
- September 2026: not recommended
- October 2026: recommended
- November 2026: not recommended (month in progress)
Why it doesn’t carry the badge
- 11 people recommend itNeeds 10 or more, met.
- 0.7 recommenders for every criticNeeds 2 to 1 or better, not met.
- Recommended in 3 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
Solves long context in architecture
“Awesome stuff. Solving context at the model architecture layer rather than trying to bolt on extra memory is the right direction IMO.”
Comment from 16 Jun 2026 in “Subquadratic – Introducing SubQ 1.1 Small” “This is pretty remarkable. We've spent a lot of time finding workarounds for LLMs reading long docs.”
Comment from 5 May 2026 in “SubQ: a sub-quadratic LLM with 12M-token context” “Looks like long context isn’t a problem anymore”
Comment from 5 May 2026 in “SubQ: a sub-quadratic LLM with 12M-token context” Faster and cheaper than transformers
“doesn’t have context rot, is orders of magnitude faster, and uses less compute that current transformer models”
Comment from 5 May 2026 in “SubQ: a sub-quadratic LLM with 12M-token context” “SubQ claims to reason as good as the frontier models, not better, but reasonably as good, and is 5x cheaper”
Comment from 19 May 2026 in “AI Is Too Expensive” Seen as architectural breakthrough
“Awesome stuff. Solving context at the model architecture layer rather than trying to bolt on extra memory is the right direction IMO.”
Comment from 16 Jun 2026 in “Subquadratic – Introducing SubQ 1.1 Small” “it's a real architectural breakthrough if true”
Comment from 5 May 2026 in “SubQ: a sub-quadratic LLM with 12M-token context” Impressive technology overall
“This is pretty remarkable. We've spent a lot of time finding workarounds for LLMs reading long docs.”
Comment from 5 May 2026 in “SubQ: a sub-quadratic LLM with 12M-token context” “I love the tech. It does amazing things.”
Comment from 16 Jun 2026 in “Subquadratic – Introducing SubQ 1.1 Small”
What people warn about
Each mark is one comment
Looks like hype or a scam
“The only purpose of this press release seems to be making investors give them more money.”
Comment from 16 Jun 2026 in “Subquadratic – Introducing SubQ 1.1 Small” “It's been all talk and no action ever since their first announcement.”
Comment from 16 Jun 2026 in “Subquadratic – Introducing SubQ 1.1 Small” “Everything in south Florida is a scam, this is just one of many”
Comment from 21 Jun 2026 in “Subquadratic – Introducing SubQ 1.1 Small” - Comment from 6 May 2026 in “SubQ: a sub-quadratic LLM with 12M-token context”
- Comment from 17 Jun 2026 in “Subquadratic – Introducing SubQ 1.1 Small”
- Comment from 6 May 2026 in “SubQ: a sub-quadratic LLM with 12M-token context”
No technical report or details
“Can you back up your claims? Why did you not release the white paper in parallel with the product? Feels really fishy.”
Comment from 6 May 2026 in “SubQ: a sub-quadratic LLM with 12M-token context” “The mechanism by which SSA meets these requirements is outside the scope of this report”
Comment from 16 Jun 2026 in “Subquadratic – Introducing SubQ 1.1 Small” “Disappointing they don't actually say how their sparse attention mechanism works.”
Comment from 16 Jun 2026 in “Subquadratic – Introducing SubQ 1.1 Small” - Comment from 16 Jun 2026 in “Subquadratic – Introducing SubQ 1.1 Small”
- Comment from 5 May 2026 in “SubQ: a sub-quadratic LLM with 12M-token context”
Lacks transparency, no independent verification
“it seems that they aren't being fully transparent”
Comment from 16 Jun 2026 in “Subquadratic – Introducing SubQ 1.1 Small” “Can you back up your claims? Why did you not release the white paper in parallel with the product? Feels really fishy.”
Comment from 6 May 2026 in “SubQ: a sub-quadratic LLM with 12M-token context” “No API access for independent verification - vaporware.”
Comment from 9 May 2026 in “SubQ: a sub-quadratic LLM with 12M-token context” Quality lags frontier models
“The delta between their score (81%) vs. Opus (87%) on SWE bench was hugely minimized”
Comment from 5 May 2026 in “SubQ: a sub-quadratic LLM with 12M-token context” “12 million token context with worse quality means this isn't going anywhere”
Comment from 11 May 2026 in “SubQ: a sub-quadratic LLM with 12M-token context”
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 |
| SubQThis page | 11 | 14 | 44% | 3 | — | — |
What people said
24 of the 27 comments that mention it, newest first. Each links to the original comment.
“longer context and image processing that has been coming out ahead in benchmarks”
“Everything in south Florida is a scam, this is just one of many”
“6450% less compute? Is Trump working there?”
“The mechanism by which SSA meets these requirements is outside the scope of this report”
“there are plenty of details that they did release that show that they don't know what they're doing”
“The only purpose of this press release seems to be making investors give them more money.”
“I love the tech. It does amazing things.”
“it seems that they aren't being fully transparent”
“It's been all talk and no action ever since their first announcement.”
“Awesome stuff. Solving context at the model architecture layer rather than trying to bolt on extra memory is the right direction IMO.”
“the lack of details makes me default not trust this”
“Disappointing they don't actually say how their sparse attention mechanism works.”
“SubQ claims to reason as good as the frontier models, not better, but reasonably as good, and is 5x cheaper”
“12 million token context with worse quality means this isn't going anywhere”
“No API access for independent verification - vaporware.”
“Can you back up your claims? Why did you not release the white paper in parallel with the product? Feels really fishy.”
“Yes, this product doesn't exist.”
“Yes you're missing something: the snake oil.”
“The delta between their score (81%) vs. Opus (87%) on SWE bench was hugely minimized”
“not launching with a technical report is a big miss”
“it's a real architectural breakthrough if true”
“doesn’t have context rot, is orders of magnitude faster, and uses less compute that current transformer models”
“Looks like long context isn’t a problem anymore”
“This is pretty remarkable. We've spent a lot of time finding workarounds for LLMs reading long docs.”
Counts come from 1,795,640 Hacker News comments, November 2025 to November 2026. How we count. Report a mistake.