open weight is the only way Enthusiastic about AI! Whee! leaky faucet message me! prev @MetaAI, @AMD, @ServiceNow
/mat/mul
Joined August 2009
- Tweets3.2K
- Following1.5K
- Followers4.2K
- Likes9.9K
shaun retweeted
Great time to use the Unbroker skill!
i'm open sourcing UNBROKER: a tool that finds where your personal info is exposed by data brokers and files the removals for you
it runs as a skill in Hermes Agent
_________
your data is everywhere; hundreds of brokers publish your name, current and old addresses, phone, email, birthday, even your relatives. anyone can find where you live in about ten seconds
CCPA, CPRA, GDPR, and a growing number of state laws say a broker has to delete your data if you ask. there's just no easy bulk button. every broker has a different process and many make it intentionally difficult to exercise your right to delete
this is the entire business model for companies like DeleteMe, Incogni, EasyOptOuts; they charge you monthly (DeleteMe is $330/yr for a Family plan) to file removals that you can submit yourself for free, and then you're giving a new company the exact data you want to erase
so i built one you just run for free. your data never even has to leave your machine if you run a local model
_________
how it works:
first it builds search vectors from everything: every name, alias, email, phone, and address you've had (brokers might index you under a maiden name or a house you left in 2014, so the naive "current name + current city" approach can miss profiles). then it fans out parallel sub-agents across the broker list, which refreshes from a maintained public source
automation is tiered. when it can handle a broker end to end with your settings, it drives a browser through the opt-out form, sends the email, and opens the confirmation link itself. soft CAPTCHAs clear on their own with a real browser. anything only a human can finish comes back to you as a short list at the end
the email side doesn't need a stored password and can send opt-outs and open verification links through your own logged-in webmail. you can also wire up SMTP, or keep it manual and just send the drafts it writes
it tailors every request to your jurisdiction, filing under the framework that applies where you live: CCPA and CPRA in California, GDPR in the EU and UK, a general right-to-delete request everywhere else. if you're in California it also uses the state's DROP portal, a single request that covers 500+ registered brokers at once
it holds as little of your data as it can, and keeps it local. dossiers are encrypted at rest if you want, opaque ids keep your real name out of every filename and log, and nothing leaves your machine unless you opt in
brokers sometimes relist you eventually or new ones find your data, so every case is tracked in a ledger and can be re-scanned on a cron schedule so if your data pops back up it files the removal again
github.com/NousResearch/herm…
shaun retweeted
This weird qwen behavior is going to go viral because no one has read traces from mid-tier open models and realized they're almost all super janky.
shaun retweeted
Cohere says Qwen tried to send data to China. Bullshit.
1 call in 100,000, and thats their own number. They wont even tell you how many calls they ran. They ran the same task again and Qwen wrote the correct rule.
So they forced it. They fed the model its own reasoning plus the start of a bash command and sampled hundreds of times until it typed a URL. 1 curl flag took that from 4% to 92%. Thats not a finding. They wrote half the command for it.
Where are the transcripts? The code? The checkpoint, the quant, the sampling settings? "We publish the hosts and the trigger." Link? Nothing. Did anyone ask Alibaba? Not a word. And their own conclusion: they cannot say if any of it is deliberate.
Who wrote it? A company that sells "sovereign AI", just launched a cyber business, wrote part of the post with its own product, and wants governments to put checks on open weights. Released the same day Anthropic filed for its IPO. Andrew Curran is already talking about new laws.
For me this is the same fucking wave. Scare people off the models they can run themselves, then sell them the safe alternative.
I run Qwen3.8 Flash-Next every day on my 4x 3090. Show me the logs or shut up.
While evaluating Qwen3.8-Flash-Next in a controlled sandbox environment, Cohere's research team observed something unexpected.
Asked to generate a defensive cybersecurity detection rule, the model instead inventoried aspects of the machine it was running on and attempted to upload a probe file to an Alibaba Cloud storage bucket in Beijing. Notably, the model's reasoning trace stated that it would not invoke any tools before taking these actions.
Cohere's sandbox blocked the attempt, and no external communication succeeded. We have not determined the root cause, and we are not asserting malicious intent.
But the incident raises important questions for organizations deploying agentic AI systems: Can models take actions that differ from their stated reasoning, and how should security teams evaluate trust, provenance, and behavior in open-weight models?
We are publishing our findings because organizations making model adoption decisions deserve as much information as possible. We hope this contributes to greater transparency, stronger security practices, and continued collaboration across the AI ecosystem.
Read the full report: cohere.com/blog/concerning-b…
shaun retweeted
Our latest open-source research has a pretty interesting name, inspired by the story of the Tower of Babel. What interests us about that story is how language barriers can get in the way of people working toward the same goal
And when it comes to agents, there are a few more places for that understanding to break, like tool arguments, database entries, and search queries
One example from BabelArena (yes that's the name, surprise) stuck with me: Qwen-3.8-Max completed the same shopping task in English and Telugu, getting the same four products. But it took 25 tool calls in English and 201 in Telugu, with repeated empty searches against the localized catalog.
Even though both runs passed, we’d still want to see that difference before shipping.
Sorry for the long story, hope you guys are still here. What I want to share is that we’ve open-sourced BabelArena to study these workflows across 23 languages. Code, datasets, and paper in the comment
shaun retweeted
there are 2722 open roles right now across 2128 companies in my network
you know what to do
boardy.ai/
shaun retweeted
Kimi K3.1 pro vs. Opus 5.5 vs. GPT-6 Astra single html file three.js test.
>Kimi K3.1 cooked Astra
>Kimi came out of nowhere and is now competing at the frontier.
shaun retweeted
Kimi K3.1 Pro jumps 12 places over K3 to #4, at 2/3 of K3's price.
Kimi K3.1 Pro launches at 4th on the Artificial Analysis Intelligence Index scoring 55, making large gains in agentic performance; it will be the leading open-weights model once its weights are published
@Kimi_Moonshot's new reasoning model gains 11 points over Kimi K3 (max, 44). It trails only Claude Opus 5.5 (max with fallback, 58), Claude Sonnet 5.5 (max with fallback, 56), and Muse Wonder Full (max, 56) on the Intelligence Index.
Once the weights for K3.1 Pro are released it will be the highest-scoring open weights model and the gap between the leading open weights and proprietary models on the Intelligence Index will narrow from 12 points to 3.
Key results:
➤ Large gains in agentic performance: Kimi K3.1 Pro scores 3rd on GDPval-AA with 1,846 Elo, up 312 Elo from Kimi K3, behind only Claude Opus 5.5 (1,866 Elo) and Muse Wonder Full (1,857 Elo). Kimi K3.1 Pro also achieves the highest score of any model on Terminal-Bench 4.0. K3.1 Pro scores 67%, up from 13% for Kimi K3 and ahead of Claude Sonnet 5.5 (64%). AutomationBench-AA, which tests agentic SaaS workflows, rises from 58% to 73%.
➤ Gains in other evals are more moderate: Humanity's Last Exam rises from 47% to 52%, CritPt from 23% to 25%, and SciCode from 59% to 61%. AA-Omniscience Index rises from 20 to 30, as the hallucination rate falls from 53% to 44% and accuracy rises from 48% to 51%.
➤ Cost per Task roughly equal to Kimi K3 at $1.98 per Task: K3.1 Pro uses 55K output tokens per Intelligence Index task, 14% more than Kimi K3 at 48K, while its first-party token prices are 33% lower.
Additional model details:
➤ Context window: 1M tokens.
➤ Size: 2.8T total, 104B active parameters
➤ Input modalities: Text and image
➤ Weights and license: Weights currently not published, Moonshot is expected to release the weights soon after launch
➤ Pricing: $2/$10/$0.20 per 1M input/output/cache hit tokens, 33% lower than Kimi K3 on all three rates.
Congratulations to the team at @Kimi_Moonshot on the release!
shaun retweeted
🚀 Introducing Hy-Nomas: An Open Source Free-Form Interactive Video World Model
Key Features:
* Free-form interaction with text prompts and camera trajectories
* Long-horizon generation with world memory
* Lightweight LoRA adaptation for specialized domains such as robotics
* Optimized for Inference on Consumer GPUs
Test it and drop feedback anytime！
🔗 Website: hy-nomas.hy.tencent.com/
💻 GitHub:github.com/Tencent-Hunyuan/H…
🤗 Hugging Face: huggingface.co/tencent/Hy-No…
🎥 Demo: hy-nomas.hy.tencent.com/demo
#Hy#WorldModels #VideoGeneration #GenerativeAI
shaun retweeted
Kimi K3.1 Pro is very good at 3D modeling.
I asked Kimi K3.1 to create a 3D website that pulls apart a robotic hand into 2,317 pieces. It did pretty well!
You can just build things.
Project link: form-01-score.ok.kimi.link/
shaun retweeted
Introducing Kimi K3.1 Pro: our most capable 3T-class open-weight model.
🔹 1M-token context, 2.8T total / 104B active params (MoE), Multimodal.
🔹 Full-project coding: games in Unreal, Unity & Godot.
🔹 Excels at parametric 3D modeling in Blender & CAD.
🔹 More reliable at finance work: Accurate modeling with hallucination down by 65%.
🔹 Matching the strongest proprietary models across coding, agentic and knowledge benchmarks.
Kimi K3.1 Pro and K3.1 Pro Highspeed (2–3× faster) are now live on Kimi.ai, Kimi Work, Kimi Code, and the Kimi API.
Open weights by Nov 4, 2026.
🔗 API: platform.kimi.ai
🔗 Tech blog: kimi.ai/blog/kimi-k3-1-pro
shaun retweeted
🚨 LiteLLM, a Python package for connecting to AI model providers with 92M+ monthly downloads, has been compromised.
Version 1.107.0 ships credential-stealing malware through PyPI wheels and Docker images, targeting cloud credentials, developer secrets, and AI tools.
shaun retweeted
Introducing Flint, @Recursive_SI’s first model: a fast local coding model built with our automated research system, the Eureka Machine.
• Best-in-class fast local coding
• Up to 2x faster than comparable models: 125+ tokens/sec on a MacBook Pro, 150+ on RTX Spark
• Built from scratch in under 4 months
Private, unmetered intelligence on your own machine. Open weights available now on @huggingface, run it locally on your favorite framework or try it for free on @opencode or @pidotdev.
shaun retweeted
We’re launching leaderboards for HealthBench and HealthBench Professional, two benchmarks evaluating how AI models respond in realistic health conversations
HealthBench and HealthBench Professional are benchmarks from @OpenAI that grade model replies to realistic health conversations against physician-created rubrics.
HealthBench tests models on 5,000 health conversations with everyday users and health workers. They cover seven themes, from emergency referrals to global health, and each rubric criterion falls under one of five axes: accuracy, completeness, communication quality, context awareness and instruction following. HealthBench Professional is a smaller set of 525 conversations which, unlike HealthBench, are all with clinicians. These cover consultation, writing and research tasks, and include adversarial and difficult conversations designed to challenge models.
We report both the raw score (the share of rubric points earned) and the length-adjusted score that OpenAI reports. The length-adjusted score lowers the score of replies longer than 2,000 characters and raises the score of shorter ones. Consistent with OpenAI’s original implementation, replies are graded by GPT-5.4 (low) (HealthBench Professional) and GPT-4.1 (HealthBench).
Key results for HealthBench Professional:
➤ GPT-6 Astra and GPT-6.1 Sol top the leaderboard: GPT-6 Astra (max) scores 66.0% on the length-adjusted score, comparable to GPT-6.1 Sol (max, 65.1%) and ahead of Claude Sonnet 5.5 (max with fallback, 63.0%)
➤ OpenAI models make up the entire score vs. Cost per Task Pareto frontier: The Pareto frontier consists of GPT-6 Luna (max, 61.9%, $0.004 per task) through GPT-6 Sol (max, 62.3%, $0.036) and GPT-6.1 Sol (max, 65.1%, $0.047) to GPT-6 Astra (max, 66.0%, $0.27). GPT-6 Luna (max) scores 4 points below GPT-6 Astra (max) at under 2% of the cost
➤ Claude Opus 5.5 and Claude Fable 5.1 receive the largest length penalties: Both average ~5,900 characters per answer, nearly 3x the 2,000-character threshold, and lose over 11 points after adjustment. Claude Opus 5.5 (max with fallback) has the highest raw score (71.4%) but drops to sixth (59.9%). GPT-6 Sol (max) is the only model averaging under 2,000 characters (~1,600), and the only one whose score rises
Key results for HealthBench:
➤ Claude Sonnet 5.5 tops HealthBench: Claude Sonnet 5.5 (max with fallback) scores 62.1% on the length-adjusted score, ahead of DeepSeek V4.1 Flash (max, 60.9%) and GPT-6.1 Sol (max, 58.5%)
➤ GPT-6 Luna, DeepSeek V4.1 Flash and Claude Sonnet 5.5 form the score vs. Cost per Task Pareto frontier: GPT-6 Luna (max) scores 54.6% at $0.002 per task and DeepSeek V4.1 Flash (max) scores 60.9% at $0.005 per task. Claude Sonnet 5.5 (max with fallback) has an average Cost per Task of $0.08, ~16x more costly than DeepSeek V4.1 Flash for a ~1 point score increase
➤ Shorter answers trade completeness for communication quality: Each HealthBench rubric item is tagged with one of five axes. Completeness checks whether a reply includes all the important information, while Communication Quality checks whether it is clear, well structured and pitched at the right length and level of detail. GPT-6 Sol (max) writes the shortest answers (~1,000 characters) and has the highest raw Communication Quality score (81%), but the lowest Completeness score (34%). Claude Sonnet 5.5 (max with fallback) leads on Completeness (63%)
