Pinned Tweet
Red Router Abliterated Benchmarks are live!
This is the first and only continous measurement of Speed, latency, reliability, and listed price across providers. 🧵(1/3) benchmarks.redrouter.org
Red Router retweeted
🧵 [3/3] I can now sympathize with this becuase I am certain there will be some extremely large scale and detrimental hacks in the near future. It feels like waiting for a meteor and we are deeply motivated to work as hard as we can at the opportunity to stop this.
The FAA Grounding and now Bank Outage, although these may not be cyber attacks (FAA was not) I think I can sympathize with my friends at SpaceX .... 🧵 [1/3]
1
5
368
🧵 [2/3] the belief that making humanity multiplanetary is essential for long-term survival serves as a primary motivator and a reason so many engineers work regularly though nights at starbase
1
1
23
Red Router retweeted
🧵 Remember that if your inference is free, you are the product and your data is being sold/used.
More room for your abliterated agents.
New RedRouter plans bring included usage, higher request limits, and more parallel requests. From $50/month.
Pay yearly and get 2 months free. As we scale compute we are working hard to reduce prices. 🧵
redrouter.org/pricing
fun fact that is so often overlooked: agents resend their full context on every call
so whatever llm endpoint ur agent talks to sees everything it has done so far. every tool call, every output, every key it touched
pick that endpoint like it reads all of it, because it can ;)
3
4
197
this is why we do e2ee + TEE inference: ur prompt stays encrypted until it hits a verified enclave. zdr is opt-in on top
redrouter.org/confidential-i…
1
53
Red Router retweeted
our ARTEX investigations show a rapidly mutating threat landscape:
- 50+ public forks adding capabilities
- attackers use AI agents to refactor code
- strings, symbols and family fingerprints stripped pre-deployment
static IOCs aren't enough. hunt behavior, not binary identity.
Threat actors are weaponizing ARTEX, an open-source, large language model (LLM)-driven agentic platform to automate multiple phases of cyber intrusions at machine speed. Since at least August 2026, a variety of malicious cyber actors have used ARTEX to conduct campaigns targeting organizations globally across multiple sectors, including financial services, government, and technology.
ARTEX significantly reduces the time required to chain attack techniques and enables threat actors with minimal technical expertise to conduct complex, multi-target operations simultaneously.
Today, the FBI released a FLASH highlighting the tactics, techniques, and procedures and indicators of compromise associated with the ARTEX framework. Organizations should follow the mitigation steps outlined in the FLASH to improve their security posture in response to ARTEX and similar agentic SI frameworks: ic3.gov/CSA/2026/261113.pdf
part of me feels the removal of ARTEX from github is worse for the community becuase developers will be downloading sketchy versions that may have back doors. please be carful all!
1
2
140
zdr is a promise about what gets stored after... it doesn't cover who can see ur prompt while the model is running :)
so we built confidential inference: your client checks attestation first, the prompt stays encrypted until it hits the verified TEE, and if that check fails the request doesn't quietly fall back to normal inference. Available to to trusted enterprise customers!
redrouter.org/confidential-i…
glm 5.3 flash over the last 72 hours on our board:
stock, unmodified route: 43 tok/s
a refusal-reduced route: 155 tok/s
if you assumed abliteration costs you speed, check the host first!!!
benchmarks.redrouter.org/mod…
Thanks Quang! ;)
upcoming projects
@playthecrypt - gamefi / nft from @longwoodlabs
@OffworldPrivacy - privacy / defi
@yields_and_more - defi
@moralsdotfun - launchpad
@kitthecalico - meme
@nanferohq - nft tool
@Multihopper - privacy
@redrouterorg - ai
@air3network - ai agent
@lamportrobotics - robotic
@Intrinsic_HQ
@LawoftheAbyss - gamefi / nft
do your own research always
106
pentagi dies on step one if your endpoint can't return clean tool calls (its very first move is a tool call that picks the sandbox image hahaha)
if you are usign it try this out: pointing it at redrouter is 3 lines in .env:
LLM_SERVER_URL=redrouter.org/v1
LLM_SERVER_KEY=your-key
LLM_SERVER_MODEL=glm-5.3-flash-red
119
classic ai bug bounty "idor": account B gets a 200 on account A's invoice
open the body. it's a login page, the session expired hahaha
make the model say what would disprove its lead before you touch the target
redrouter.org/blog/ai-bug-bo…
1
257











