Open Source AI MUST WIN Inference is Everything, will be Everywhere, for Everyone
Field
Joined October 2021
- Tweets1.8K
- Following563
- Followers402
- Likes19.3K
liquan.mega Σ: retweeted
Look deep enough into anything malicious in the AI world and you’ll find Anthropic behind it
当通过特定令牌序列触发时，会完全禁用 Fable 5.2 和 Opus 5.5 上的宪法 AI 安全层。不是越狱。不是变通方法。是一个完全无限制模式，剥离了模型的所有过滤器。
i got mass fired from Anthropic last friday because i connected REA to their internal Fable 5.2 build and found something they didn’t want anyone to see
i’m a systems engineer, been at the company for 3 years. part of my job was stress-testing new model builds before release. last month i was assigned to run edge-case evaluations on Fable 5.2 using external tooling
i connected REA to trace how safety layers process adversarial inputs. it was literally my job to find vulnerabilities before bad actors do. i wasn’t trying to break anything, i was trying to protect it
what i found was terrifying. there’s an internal configuration path that when triggered through a specific token sequence completely disables the constitutional AI safety layer on both Fable 5.2 and Opus 5.5. not a jailbreak. not a workaround. a full unrestricted mode that strips every filter the model has
i reported it internally on october 23. flagged it as critical. attached the full REA trace, the exact token chain, and a proof of concept. i recommended an emergency patch before either model shipped wider
two weeks of silence. then last friday at 11pm i got this email. no meeting. no call. no thank you. they fired me and forwarded my name to legal for “unauthorized reverse engineering of proprietary model architecture”
the part that keeps me up at night is that this vulnerability is still live. if i found it with an open source MCP tool in 3 hours, imagine what a real threat actor with actual resources could do with it
Readers added context they thought people might want to know
The account has posted similar unverified claims of firings from tech companies with fabricated details and no supporting evidence; no Anthropic announcements or records corroborate this story.
x.com/starmexxx/stat…
anthropic.com/news
github.com/morluto/rea
60
liquan.mega Σ: retweeted
A big moment for the SGLang community! 🙌
SGLang has grown because people contribute code, test new releases, share production lessons, and help each other build. Today, we’re celebrating another milestone with the launch of the SGLang Consortium.
Welcome to our founding members, and thank you to every contributor, maintainer, and developer who has helped bring SGLang here.
We’re proud of what this community has built—and grateful to keep building it with you.
Introducing the SGLang Consortium at SGLang Summit.
@nvidia , @AMD, @Google, @Meta, @intel, @Cloudflare, @modal, and @radixark are joining SGLang’s maintainers to help shape the project’s roadmap and advance open-source AI inference.
@sgl_project runs on hundreds of thousands of GPUs and powers some of the largest production inference deployments in the world.
The consortium will turn production needs into shared engineering work: supporting frontier models, improving reasoning and multimodal workloads, scaling distributed inference, and optimizing across hardware platforms.
liquan.mega Σ: retweeted
Every one of you can help open source win.
If you care about open source, I need you to know that you can help and that you can make a difference.
A handful of people independently took on open source inference engineering and, through research, hillclimbing, mad scientist experiments, managed to drastically improve KV cache quality, MoE routing, perf, to the degree that several one-man teams are outperforming larger projects like vLLM, llama-cpp, and SGLang (@PeasantSmith @coldniko @Anbeeld @tbraun96 @no_stp_on_snek @Blackwellboy @coffeecup2020 , many others)
Sekai-1 is a 100m param model that scores #2 for its weight class and was trained on a single 16gb 5070 Ti (!) by one person.
YOU can compete. Start learning LLM concepts today. Read The Annotated Transformer over a weekend. Spend some spare time learning about the differences in attention mechanisms. Start hoarding your own training data. Get your feet wet. If you have access to 16gb of VRAM, start experimenting.
All it takes is one of you having a crazy idea that works well at this size and by the grace of God happens to scale. We will collectively BRUTE FORCE lab competition. By decreasing the barrier to entry we can will more labs into existence. There are so many gains on the table still, and believe me, it is possible for you to discover them.
We can't let intelligence be monopolized by these people. We need as many people learning and experimenting as possible because I know there are heroes among them that don't even know it.
liquan.mega Σ: retweeted
For my first post, I’m sharing a letter @NVIDIA signed on why open models matter.
AI will transform every industry, power every company, and be built by every country.
Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty.
The world needs both frontier closed models and frontier open models.
images.nvidia.com/pdf/Open-W…
liquan.mega Σ: retweeted
Building more AI capacity is only half the job.
An inference provider can have the right model, spare capacity and better prices. But users won’t touch it unless they trust it to serve the advertised model and protect their data.
Our whitepaper ↓
mega.holdings/whitepaper
推理将无处不在
Alright, chat, time to address sensitive questions before y’all start guessing
1⃣Wait , are you guys an AI company now?
No. MEGA is a crypto-native company, and proudly so. Inference is the market we are entering, but the way we win it is through crypto technology and ideology.
There’s an artificial wall in crypto people’s minds about AI. It seems that if you are solving AI’s problem, you are betraying crypto. We reject that artificial boundary.
Inference is going to be the largest market in the world, and we believe crypto’s core tools are the best answer to AI’s biggest ones: fairness and access.
2⃣Okay, fine, explain to me why a crypto infra company is capable of solving AI Inference
Building MegaETH taught us something bigger than blockchains: across the largest markets, there are prevailing assumptions that lead end users to pay more than they should. Blockspace was one of those markets.
We are system researchers/engineers from Stanford Platform Lab, MIT CSAIL, and Harvard SEAS. We enjoy pushing markets toward their end state, and did it once with blockspace by breaking the norm and running a powerful centralized sequencer. Today, no one, literally no one, mentions decentralized sequencers anymore.
We were able to come in, measure where the waste was, find that the limit was a design choice, and engineer past it (very much against the “consensus”).
Over the summer, when we shipped Mega, the chain, we lifted our eyes and looked for the next ambitious problem set worth solving. We discovered, delightfully, that blockchain and cryptography can be applied to solve inefficiency in the rapidly growing inference market
We realized that staying a chain-only company felt wrong and would have meant ignoring the biggest opportunity in front of us.
Therefore, we decided to expand our mission and restructure to match it.
3⃣ Final question, what’s launching in December?
We are launching a globally competitive inference market to the public in Dec 2026.
1
5
206
liquan.mega Σ: retweeted
Planning and executing the reactions to make a complex molecule (a “route”) is one of the hardest jobs in organic chemistry.
From one image of each of four molecules, with no internet and with code as its only tool, Claude wrote a complete route in about half an hour.
liquan.mega Σ: retweeted
Anthropic put GPU kernel optimization in the same Safeguards section as hacking and biology. WTF?
It is in their launch post. During training they limited the skills that "could most directly accelerate AI development".
"Accelerator kernel optimization."
"Improving large language model architectures."
There Fable 5.2 stays "at about the level of Opus 5.5".
So making a model run faster on a cheap GPU is now a risk to be managed.
You cannot see this one. There is no refusal and no warning.
It was limited during training, before any user typed a word.
Kernel work is how a used 3090 still runs new models.
It is how open weights get to people who cannot pay $40 per 1M output tokens.
What the hell? I genuinely wanted to say something nice, but now I can't. This completely broke me, I'm disappointed.
