Share this page!

Enter URL or ID to Unroll

You can paste full URL like: https://x.com/threadreaderapp/status/1644127596119195649
or just the ID like: 1644127596119195649

How to get URL link on X (Twitter) App

  1. On the Twitter thread, click on or icon on the bottom
  2. Click again on or Share Via icon
  3. Click on Copy Link to Tweet
  4. Paste it above and click "Unroll Thread"!
  5. More info at Twitter Help
Ankesh Anand Profile picture
@ankesh_anand
Dec 19, 2025 • 1 tweets • 1 min read • Read on X
"how can flash beat pro??" -> the answer is RL!

flash is not just a distilled pro. we've had lots of exciting research progress on agentic RL which made its way into flash but was too late for pro.

can't wait to finally bring them to pro👀

• • •

Missing some Tweet in this thread? You can try to force a refresh
　

Keep Current with Ankesh Anand

Ankesh Anand Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @ankesh_anand

Ankesh Anand Profile picture
Jan 29, 2025
The DeepSeek discourse is simultaneously under-crediting and over-crediting them for what they achieved. So, some quick thoughts:
1. Re Distillation claims: DeepSeekCoder-V2 [1] was released in June 2024, and they had RL on verifiable rewards working back then with great success. They were the only team outside of Gemini and OpenAI I knew of that were RL-pilled.

6 months from then on, I place a very low prior this team would just distill from o1 given how much info about o1 was in the OpenAI blog post itself. They may have used o1 CoTs as examples to seed their human CoT data, but again the RL team there is talented enough to do it on their own.
[1] arxiv.org/abs/2406.11931Image
2. The 5.5M$ thing is entirely believable for "one final run", and a massive feat. It is however only surprising if you compare with llama3 training costs which was a couple generations behind as Wenfeng mentions here. Obviously, the total R&D costs are much higher. Image
Read 5 tweets

Did Thread Reader help you today?

Support us! We are indie developers!

This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us!

Send Email!

:(