←
ⓧ Richard DeVaul 🚀🇺🇸💪 devaul.sol4,503 posts
ⓧ Richard DeVaul 🚀🇺🇸💪 devaul.sol avatar

ⓧ Richard DeVaul 🚀🇺🇸💪 devaul.sol

@rdevaul

Head of AWERXAI at @relativityspace. Fomer Co-founder @XNet_Mobile ⓧ, Former CTO Google X. @rdevaul on LinkedIn, GitHub, writing on Medium

◷ Joined December 2008▧ 366 Photos and videos
ⓧ Richard DeVaul 🚀🇺🇸💪 devaul.sol avatar
ⓧ Richard DeVaul 🚀🇺🇸💪 devaul.sol@rdevaul·Nov 4

If these performance numbers hold up in real world use, this could be a game changer. As an aerospace professional, I don’t want to have to choose between high-performance open weights and US based for my agentic engineering applications.

Cruz@ViC305Nov 4

🚨 At last a USA based company who invested $32M in compute to train this model from scratch! They beat Qwen3.8-27B and 35 B-A3B and all the leading USA frontier models like Beam 501B and Meta Muse Spark Open 816B on multiple benchmarks! 🚀🔥🤯 Ripple-65B Preview (65B dense) is punching well above its size on agentic coding: • TerminalBench 2.1: 80.5% • SWE-Bench Pro v2.0-Hard: 79.1% • DeepSWE v1.1: 44.3% For context vs smaller models: Qwen3.8-27B → 73.0% / 61.7% / 42.2% Qwen3.6-35B-A3B → ~51.5% (TB 2.0) / 49.5% As per their read me “a 65B parameter dense Transformer model pretrained on 12 trillion tokens with support for up to 1M context length. Ripple-65B Preview delivers strong results on agentic coding, scoring 80.52 on TerminalBench 2.1 and surpassing significantly-larger US open-weight models such as Reflection Beam 501B (80.1), Meta Muse Spark Open 816B (76.4), Thinking Machines Inkling 975B (63.8), and NVIDIA Nemotron 3 Ultra 550B (56.4).” 🤯🤯 Going to put this model through its paces and really validate these claims! Stay tuned!!

Cruz avatar
⟳ ⓧ Richard DeVaul 🚀🇺🇸💪 devaul.sol retweeted
Cruz@ViC305·Nov 3

It’s extremely important that the open sourced community is made aware of all quantization formats that can provide the most value for the least amount of GPU/VRAM and that’s where EXL3 by @turboderp_ comes in to shine. I’ve coined my own EXL3 quantization process called SAGE(Sentinel Assisted GPU-aware EXL3) that has grown over the last few months into a contender that can score higher than even its 8bit GGUF counter parts with Deepseek v4.1 Flash 1017 being the model in the spotlight, while running on one DGX Spark. I’ve been able to achieve a 94% top agreement at 1.55Bpw effectively breaking into the One DGX Spark Pareto Frontier. With that being said, here’s an article of proof that I, a one man team with the resources of a company I work for that’s currently in stealth, put together to demonstrate the strengths and weaknesses of my quantization process on how a sub 2bit quant can exceed the intelligence and quality of 2-8bit quantizations in GGUF. You’ll also find a HuggingFace dataset of proofs and results! Thank you to @UnslothAI & @mhelcig for bringing up this unsolicited challenge by pushing the capabilities of GGUF forward and @NVIDIAAI @NVIDIARTXSpark for bringing frontier model access to our desk by ways of the RTX and DGX Spark platforms! Here’s to fitting more of what counts in the GB10 🫡🙏 🏆 HMMT: 70.27% above all 8. 🧮 AIME: 95.42% above 5/8. 📋 IFEval: 92.05% above 1/8. Article: x.com/vic305/status/21173776…

0:10

We benchmarked DeepSeek-V4.1-Flash-1017 GGUFs on Terminal Bench, DeepSWE & more on 512K context. Unsloth ranks highest overall. 1-bit Unsloth GGUF scores 68.1 on DeepSWE, 72.2 on Terminal-Bench, 52.7 on AutomationBench. >10% ahead of other quants. GGUF: huggingface.co/unsloth/DeepS…

Post attachment
ⓧ Richard DeVaul 🚀🇺🇸💪 devaul.sol avatar
ⓧ Richard DeVaul 🚀🇺🇸💪 devaul.sol@rdevaul·Oct 23

Heh - I appreciate the github.com/rdevaul/yapCAD shout-out, Matt :)

left Claude/Codex alone with a 6 hour deadline to make "unhinged" promo for my agent-portal tool

3:26
MattyG avatar
⟳ ⓧ Richard DeVaul 🚀🇺🇸💪 devaul.sol retweeted
MattyG@meawoppl·Oct 23

left Claude/Codex alone with a 6 hour deadline to make "unhinged" promo for my agent-portal tool

3:26
ⓧ Richard DeVaul 🚀🇺🇸💪 devaul.sol avatar
ⓧ Richard DeVaul 🚀🇺🇸💪 devaul.sol@rdevaul·Oct 23

Do you think the current "bug tsunami" is bad? I just did back-of-the-envelope math and estimate the number of the number of critical, undiscovered vulnerabilities in widely-deployed OSS projects is conservatively 2,000 to 10,000, with 100-300 "PhP grade" - get ready

ⓧ Richard DeVaul 🚀🇺🇸💪 devaul.sol avatar
ⓧ Richard DeVaul 🚀🇺🇸💪 devaul.sol@rdevaul·Oct 21

To be fair, there are things about OpenShell that I like, and important features that xshell has designed that I think OpenShell is lacking. But it's good to see that agent accountability and reliablity are now first-class concerns for production workflows

ⓧ Richard DeVaul 🚀🇺🇸💪 devaul.sol avatar
ⓧ Richard DeVaul 🚀🇺🇸💪 devaul.sol@rdevaul·Oct 19

proving that you can print and build the github.com/rdevaul/yapRover — grab and agent and build your own project with 100% open-source agent-first advanced CAD: github.com/rdevaul/yapCAD

We’re not just vibe coding hardware, we’re using agents to take hardware through the entire design for manufacturing loop

Post attachmentPost attachment