Ricursive Intelligence

[Ricursive Intelligence @RicursiveAI]

Today we release Ripple-65B Preview, our 65B dense LLM we pretrained from scratch.

Ripple sets a new state-of-the-art for U.S. open-weight models in agentic coding tasks, surpassing significantly larger models including Muse Spark Open-816B, Beam-501B, Inkling-975B, and Nemotron 3 Ultra-550B.

Learn more: https://t.co/Pe04ZKhkE9

[Ricursive Intelligence @RicursiveAI]

Last week, we released Ripple-65B Preview. A 65B dense LLM that set a new U.S. open-weight SOTA on agentic coding.

Join our very own @DZhang50, Founding Member of Technical Staff, and @togethercompute's @zainhas to learn how we did it.

📆 Tuesday, Nov 17 · 11am PT / 2pm ET
👉 Register at https://t.co/d4OUjY9L4E

[Together AI @togethercompute]

How do you build a small coding model from scratch?

@RicursiveAI 's Dan Zhang breaks down the architecture and decisions behind Ripple-65B Preview with @zainhas: dense over MoE, the ablations that held up (and didn’t), and what happened inside the RL runs.

Tuesday, Nov 17 . 11 A.M. PT / 2 P.M. ET

[Ricursive Intelligence @RicursiveAI]

Ripple-65B Preview is now available on ollama!

https://t.co/rbUSgnWH80

[ollama @ollama]

Ripple 65-B Preview from @RicursiveAI is on Ollama. Built for agentic coding, with up to 1M tokens of context. Run it locally with full zero data retention.

[Ricursive Intelligence @RicursiveAI]

Last week we released Ripple-65B Preview. This Friday, come build with it.

Ricursive After Dark Hackathon: Fri Nov 13, 3PM–10PM, at our Palo Alto office. Compute, dinner, and hardware prizes.

Applications close Thursday, November 12 at noon PT. Accepted participants will be notified by email Thursday night with the address and full details.

https://t.co/zkOcwkS8nD

[Ricursive Intelligence @RicursiveAI]

@shaunmmaguire Thanks, Shaun!

[Ricursive Intelligence @RicursiveAI]

Today we release Ripple-65B Preview, our 65B dense LLM we pretrained from scratch.

Ripple sets a new state-of-the-art for U.S. open-weight models in agentic coding tasks, surpassing significantly larger models including Muse Spark Open-816B, Beam-501B, Inkling-975B, and Nemotron 3 Ultra-550B.

Learn more: https://t.co/Pe04ZKhkE9

[Ricursive Intelligence @RicursiveAI]

Apply here: https://t.co/xBlqknJvJO

Read about how we built Ripple-65B Preview: https://t.co/JOebbvGHtQ

[Ricursive Intelligence @RicursiveAI]

Now live on Fireworks and Together as well! All inference endpoints available below:

Fireworks: https://t.co/ynpk146es2
Together: https://t.co/C8d48YBPkM
GMI Cloud: https://t.co/aGiqARWjYh
OpenRouter: https://t.co/Hq7JbJgDun

[Ebrahim Songhori @esonghori2]

Very good and detailed review of our model from @Sam_Witteveen

Ripple 65B: A Chip Design Lab Just Beat Every US Open Model https://t.co/UMBdbzG0t9

[Ricursive Intelligence @RicursiveAI]

@Alfred_Lin Thank you, Alfred!

[Ricursive Intelligence @RicursiveAI]

@gmi_cloud @OpenRouter @huggingface @togethercompute @FireworksAI_HQ Update: the OpenRouter link is here: https://t.co/Hq7JbJgDun

[Ricursive Intelligence @RicursiveAI]

@gmi_cloud @OpenRouter @huggingface @togethercompute @FireworksAI_HQ Update: the OpenRouter link is here: https://t.co/Hq7JbJgDun

[Ricursive Intelligence @RicursiveAI]

@gmi_cloud @OpenRouter @huggingface @togethercompute @FireworksAI_HQ

[Ricursive Intelligence @RicursiveAI]

Thank you for your kind words!

Yes, our $32M R&D cost fully encompasses our pretraining ladder development, data mixing, evals, debugging, synthetic data generation, posttraining recipe development, and the final hero pre, mid, and posttraining run!

[swyx @swyx]

> we developed and trained Ripple-65B Preview for only $32M, which is one-tenth the cost of the previous open-weight SOTA on agentic coding¹

this is like the avengers of open model SOTA

¹ "Note that $32M is our full R&D cost starting from scratch, including pretraining ladder development, data mixing, evals, debugging, synthetic data generation, posttraining recipe development, and the final hero pretraining and posttraining run."

[Ricursive Intelligence @RicursiveAI]

Thanks for a great partnership @nvidia!

[NVIDIA RTX Spark @NVIDIARTXSpark]

Congrats to @RicursiveAI on Ripple-65B Preview, their first open model 🎉

It was trained on NVIDIA B300 GPUs & sharpened with open NVIDIA Nemotron post-training datasets.

Download the NVFP4 checkpoint from Hugging Face and run locally on DGX Spark 👉 https://t.co/MyEPwPTKIt

[Ricursive Intelligence @RicursiveAI]

@AcousimHss yes it does!

[Ricursive Intelligence @RicursiveAI]

@artxnik nah maybe next time

[Ricursive Intelligence @RicursiveAI]

@dodecagonism Qwen 3.8 27B: 73% TB2.1
Ripple-65B: 80.5% TB2.1

[Ricursive Intelligence @RicursiveAI]

@ukrroot we use Terminus 2 with 128K sequence length for TB2.1, the other models are their own reported numbers typically with custom harnesses and longer sequence lengths

[Ricursive Intelligence @RicursiveAI]

@torisetxd our NVFP4 checkpoint currently averages ~40 tok/s on DGX Spark, with more work it can be even better!