autoresearch
No written verdict yet. Here is what commenters said.
8recommend
1warn against
Mar 8, 2026first mention
Sep 19, 2026latest mention
Why people recommend it
“which is extremely powerful when applied to problems that generalize well to it”
Show 3 more
“Seemed like that one was safe”
“autoresearch managed to improve the harness performance”
Caveats and dissent
“likely not worth the $$$ spent on the tokens”
Often compared with
From the makers not counted in the ranking
“My fork of Andrej's repo if you want to join in”
Threads where it came up
- I tried Karpathy's Autoresearch on an old research project 3 mentions
- Science Is Open Software 1 mention
- DeepSeek V4 Pro 0813 1 mention
- Discovery Loop 1 mention
- Show HN: DOM-docx – HTML to native, editable Word docs (MIT) 1 mention
- Guy beats one of the best published quantum circuits for breaking ECDSA with AI 1 mention
Also written as Autoresearch, AutoResearch. Data: autoresearch.json. Quotes are verbatim; the verdict and theme labels are written by a model from these comments, and every clause links to its sources.
All comments 20
“This seems strikingly close to autoresearch”
“maybe this?”
“hard to imagine they were not influenced by autoresearch.”
Thanks, and just to mention, I’ve had awesome results using Autoresearch loops on other things like SQL performance. Prompt example: You are a SQL performance researcher. Run the following SQL query to establish a baseline, then come up with hypotheses to improve performance. Score each result and run 5 iterations. Avoid any regressions, each result must contain the exact same rows and columns. See https://github.com
“which is extremely powerful when applied to problems that generalize well to it”
“This is already known to work.”
“the most viral lately is https://github.com/karpathy/autoresearch”
“Seemed like that one was safe”
“In case you are unfamiliar with Karpathy's Loop[1]”
“Experimented with auto-research for prompt fine-tuning”
“yet it's but a highly simplified version of AlphaEvolve”
“The difference is autoresearch's val_bpb measures what the agent built”
“autoresearch managed to improve the harness performance”
“Autoresearch is nothing new, big players are already in the game with more sophisticated solutions”
“the ability to say: "optimize this benchmark and commit only if you pass these tests" is nice”
“likely not worth the $$$ spent on the tokens”
“Take a look at the GitHub repo: "forked from karpathy/autoresearch"”
“great find, create a claude code plugin”
“worth having a look at the changes the LLM did apply”
“when I had the model write up the results for that session”