frontpage.
newsnewestaskshowjobs

Made with ♥ by @iamnishanth

Open Source @Github

fp.

Open in hackernews

Tinker by Thinking Machines

https://thinkingmachines.ai/tinker/
39•ashvardanian•1h ago

Comments

ChrisArchitect•1h ago
Blog post: https://news.ycombinator.com/item?id=45441219
sh3rl0ck•54m ago
They mention their "cookbooks" but I couldn't find them... Their blog was immensely interesting so I could see this being a good entrypoint
danobi•52m ago
https://github.com/thinking-machines-lab/tinker-cookbook
neilv•52m ago
There was a famous tech company in supercomputing and AI, called "Thinking Machines", worked at by people such as Danny Hillis and even Richard Feynman.

Does this new company have some connection to that, such as some of the same people?

https://en.wikipedia.org/wiki/Thinking_Machines_Corporation

palmotea•39m ago
> Does this new company have some connection to that, such as some of the same people?

Really doubt it. This "Thinking Machines" seems like an on-trend SV AI software startup, that "Thinking Machines" was a Massachusetts-based hardware company that went bankrupt 30 years ago (and whatever's left is Oracle). That "Thinking Machines" people are retiring by now.

I wonder if Sun/Oracle let the trademark lapse, because if it's still active I'd imagine this startup's gonna get sued.

jsnell•38m ago
No relation AFAIK, this is Mira Murati's post-OpenAI joint.

Which makes the hero image on this page being a diagram of Danny Hillis' TinkerToy computer all the more baffling.

staticshock•35m ago
nope, this is Mira Murati's thinking machines. she used to be the CTO of OpenAI, but started her own thing once she [I think] realized that she was missing out on the gold rush by staying there.
tinyhouse•47m ago
This is great as we need more solutions that help people train and fine-tune models. There is a lot of open source and some companies behind some of the popular open source packages, like Unsloth, but the more the merrier, esp given that Thinking Machines has the expertise and resources to build something that last.
NitpickLawyer•32m ago
Yeah, I'm really curious about their stacked multi-tenant lora training at the same time. If this gets commoditised enough, it could be interesting to try "end of the day fine-tunes on daily conversations" and see where that leads. Or a targeted RL on "missed / rejected tasks" for an agent, after you get enough samples for a run, and so on.
next_xibalba•30m ago
This company was most recently valued at $12 billion. Tell me there's not an AI bubble.

https://www.reuters.com/technology/mira-muratis-ai-startup-t...

NitpickLawyer•23m ago
They seem to position themselves to commoditise renting shovels to use with your own data. Seems pretty bubble-safe to me.
wmf•11m ago
That's a fair valuation if they have a >10% chance of developing a frontier model. It's an excessive valuation if they just release tools to create LoRAs and such.
closewith•27m ago
The name chosen is an antiquated ethnic slur in much of the Anglosphere.
fxtentacle•13m ago
"the Tinker API provides simple functions to compute gradients, update the weights, and sample outputs from the trained model"

It sure sounds like a PyTorch tutorial, but I believe it's yet another "AI training made slightly easier for you" start-up. But all of them seem to solve the easy problem of managing data and compute, while very few tackle the hard problem of generating good training data.

HieraSynth: A Parallel Framework for Complete Super-Optimization [pdf]

https://lsrcz.github.io/files/OOPSLA25.pdf
1•matt_d•1m ago•0 comments

Mom Test August 2025 Provisional Conclusions

https://theheartattackdiet.substack.com/p/mom-test-august-2025-provisional
1•paulpauper•1m ago•0 comments

A Field Guide to Writing Styles

https://linch.substack.com/p/on-writing-styles
1•paulpauper•1m ago•0 comments

Collegetown midrise gets green light from Planning Board

https://ithacavoice.org/2025/09/collegetown-midrise-gets-green-light-from-planning-board/
1•PaulHoule•2m ago•0 comments

Brave browser passes 100M monthly active users

https://brave.com/blog/100m-mau/
1•LorenDB•3m ago•0 comments

Book Club Scams Are a Warning of Emerging AI Super-Scams

https://www.patreon.com/posts/genre-grapevine-139940838
1•ilamont•4m ago•0 comments

The Rise of Boring Scams: How AI Is Killing the Nigerian Prince

https://gregrossmunro.com/the-rise-of-boring-scams-ai-is-killing-the-nigerian-prince/
1•gregdrm•4m ago•1 comments

Hunter S. Thompson's death to be reviewed 20 years after ruled a suicide

https://www.thejournal.ie/hunter-s-thompson-death-review-suicide-6832317-Oct2025/
1•austinallegro•4m ago•0 comments

Clues from deep sea that exploding star sent debris to the Earth 10M years ago

https://phys.org/news/2025-09-clues-deep-sea-star-debris.html
1•rbanffy•5m ago•0 comments

Gboard Dial Version

https://github.com/google/mozc-devices/tree/main/mozc-dial
1•xnx•5m ago•0 comments

Electroflow: Bolstering our Lithium supply without building new mines

https://blog.usv.com/electroflow
1•wslh•6m ago•0 comments

Show HN: I over engineered changing Azure tags?

https://tagfixer.com/
1•Locrin•7m ago•1 comments

Behold: The Lego Game Boy has been modded to play games for real

https://www.theverge.com/news/790093/lego-game-boy-natalie-the-nerd
1•m-watson•7m ago•0 comments

Show HN: Summarise voice notes and YouTube links inside WhatsApp and Telegram

https://voicenxt.com
1•oathly-dev•7m ago•0 comments

OpenAI won't say whose content trained its video tool. We found some clues.

https://www.washingtonpost.com/technology/interactive/2025/openai-training-data-sora/
3•kgwgk•9m ago•0 comments

Ace Tone

https://en.wikipedia.org/wiki/Ace_Tone
4•jethronethro•9m ago•0 comments

Google Finance (Beta)

https://www.google.com/finance/
2•kirillzubovsky•9m ago•1 comments

AROS Vision – Modern and Free Amiga-Compatible Experience on Amiga and PC

https://aros-vision.de/
1•gjvc•9m ago•0 comments

Scientists make embryos from human skin DNA for first time

https://www.bbc.com/news/articles/c4g2vyee0zlo
2•bookofjoe•12m ago•0 comments

What Counts as an Iteration?

https://www.autodidacts.io/iteration/
1•Curiositry•13m ago•0 comments

How AI Demand Is Driving a Multi-Year Memory Supercycle

https://www.viksnewsletter.com/p/how-ai-demand-is-driving-a-memory-supercycle
1•klelatti•14m ago•0 comments

AirPods Pro 3 Impossible to Repair, Earn Score of 0 in iFixit Teardown

https://www.macrumors.com/2025/10/01/ifixit-airpods-pro-3-teardown/
3•LorenDB•14m ago•0 comments

DeepMind's Dreamer 4: Training Agents Inside of Scalable World Models

https://danijar.com/project/dreamer4/
1•danijar•15m ago•0 comments

The Anti-Truth Mandate: A Technical Flaw in Claude's Safety Architecture

https://github.com/anthropics/claude-code/issues/8683
4•RecursionLab•15m ago•3 comments

Grab's Migration from Go to Rust Cut Costs by 70%

https://blog.bytebytego.com/p/how-grabs-migration-from-go-to-rust
2•josephcsible•16m ago•0 comments

OpenAI on losing track in German copyright case

https://aifray.com/breaking-openai-on-losing-track-in-german-copyright-case-brought-by-music-righ...
2•smartmic•17m ago•0 comments

The Campaign for Vertical Television

https://interconnected.org/home/2025/10/01/vertical
2•ingve•17m ago•1 comments

Feedback Triggers Different Types of Perfectionism

https://journals.aom.org/doi/10.5465/amj.2023.0187.summary
1•rustoo•20m ago•0 comments

24/7 AI Livestream

https://dmwithme.com
2•0xlogk•21m ago•1 comments

Show HN: TraceML, a tool to trace live memory usage in PyTorch training

https://github.com/traceopt-ai/traceml
1•traceopt-ai•23m ago•0 comments