AutoThink
AI
reasoning optimization
SHOW HN2025
How HN took it
10 positive, 8 neutral or asking, 1 mixed, 4 critical, out of 23 comments about the project. Thread score +0.29.
Idea: 9 for, 1 againstPerformance: 1 for, 3 against
The thread, comment by comment
Each row is a top-level comment and its replies, in HN's order. Select a cell to open that comment on HN.
positive neutral or mixed critical about another commenter or something else the maker
What HN said
What people liked
Obvious optimisation, good to see it written up
From 1 comment
This is an obvious optimisation. Surprised this isn't been done already. Good job writing it up and showing how it can be done.
nssnsjsjsjs on HN
Interesting work toward competitive open reasoning models
From 1 comment
> I think the same goes for o3. Definitely, in my experience. Elsewhere in the thread, OP says that open models/systems don't do this, in which case this seems like important…
CharlesW on HN
Objections
Classifying difficulty by token count can misjudge hard questions and trade correctness for speed
From 2 comments
But how do you classify a question as high vs low complexity? Some seemingly simple questions can turn out to be very very complex. For example, integer solution to ``` x³ + y³ +…
bufferoverflow on HNSo it can be very very wrong. You're trading correctness for speed.
bufferoverflow on HN
Benchmark setup differs from official results (GPQA-Diamond 33.8 reported)
From 2 comments
Official https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B says GPQA-Diamond is 33.8
lostmsu on HNGPQA-Diamond is 200 questions. Any GPU since 2019 with 12GB of VRAM should be able to run tens if not hundreds of queries for a 1.5B model in parallel.
lostmsu on HN
Requests
Explain how the complexity classifier works and generalizes
From 2 comments
Hey, this is really interesting. What are the features you used to measure the reasoning complexity? In other words, how does one evaluate a query during classification?
Dowwie on HNIf you ask it to rethink the problem again because you've found a flaw, does it bump up the complexity and actually think about it. Like a person might give you a quick answer to…
VagabundoP on HN
Support for llama.cpp
From 1 comment
Hi, it does not work with llama.cpp right?
shirman on HN
Questions
How is query complexity measured during classification?
From 2 comments
Hey, this is really interesting. What are the features you used to measure the reasoning complexity? In other words, how does one evaluate a query during classification?
Dowwie on HNBut how do you classify a question as high vs low complexity? Some seemingly simple questions can turn out to be very very complex. For example, integer solution to ``` x³ + y³ +…
bufferoverflow on HN
Does it help local users who want maximum accuracy?
From 1 comment
If host models for others, then sure, I'm happy to save some computation time for really simple queries. Sure the cost is that the model will be effectively dismissive of…
vintermann on HN
Alternatives named
Gemini 2.5 Pro and o3 already vary thinking time
From 2 comments
> they spend the same amount of "thinking time" on "what's 2+2?" as they do on complex mathematical proofs. Not anymore. Have you seen Gemini 2.5 Pro? Ask it simple questions and…
behnamoh on HNYou didn’t invent this. Models like o3 already do it, that’s why the amount of thinking time varies.
MagicMoonlight on HN
Earlier autothink POC using LLM-assigned complexity score
From 1 comment
I, too, built a POC autothink shortly after the Claude 3.7 release that included the `extended thinking` toggle. It's literally also called autothink:…
NiloCK on HN