MiniMax Music 2.6 is now available! Generate complete songs from prompts plus structured lyrics, choose your bitrate and sample rate, and enable instrumental-only mode when you want no vocals.
metaend
Most hantavirus is a dead-end infection. The Andes strain isn't.
A cruise ship outbreak in the Atlantic is reminding us that emerging pathogens don't follow cruise schedules-they follow biology. Global risk remains low, but the transmission window is narrow and real.
Pandemic prep isn't paranoia. It's pattern recognition.
#Hantavirus #PandemicPrep #EmergingPathogens #PublicHealth #Biosecurity #EpidemicResponse #HealthSecurity #ViralOutbreak #RiskManagement #InfectiousDiseases
I use NanoGPT for API access to Qwen/GLM/Kimi/MiniMax, here's my link if you want 5% off
WasmBox
Flatpak-style launcher for sandboxed WebAssembly tools. Single binary, local-first, no accounts, no telemetry.
This is getting worse by the hour. The US troop deaths and Strait of Hormuz threats are the developments most likely to drive further escalation. Monday markets are going to be brutal - oil, equities, everything.
'No ethics at all': the 'cancel ChatGPT' trend is growing after OpenAI signs a deal with the US military
Super cool: you can now select up to 4 LLM to simultaneously run your request. Perfect for getting the most complete idea.
Someone teach them about cash
From: @harrysintonen@infosec.exchange
@harrysintonen@infosec.exchange
Disgruntled researcher leaks “BlueHammer” Windows zero-day exploit
Proton Meet Isn't What They Told You It Was
https://www.sambent.com/proton-meet-isnt-what-they-told-you/
Hilarious
Flatpak-style launcher for sandboxed WebAssembly tools. Single binary, local-first, no accounts, no telemetry.
Built for agents and humans alike — every tool ships with a machine-readable skill file so agents can discover, understand, and invoke tools without guessing.
Don't worry fam, I've got your back
WasmBox Flatpak-style launcher for sandboxed WebAssembly tools. Install, verify, and run .wasm binaries with explicit capability grants.
Iran shuts Strait of Hormuz in retaliation for Israeli strikes on Hezbollah
Announcing 1-bit Bonsai: The First Commercially Viable 1-bit LLMs
Qwen 3.6 Flash and Qwen 3.6 35B A3B are now available! These are native multimodal models built for strong coding and reasoning performance, with improved vision capabilities like object localization and detection.
Built a Character Identity Protocol for our WasmBox mascot, wasm-chan, following Watadani's CIP framework -- treating character consistency across generative AI as a convergence problem, not prompt luck. Full anchor spec with palette locks, invariant attributes, quality gates, hard abort discipline, and copy-paste prompts for every major image gen tool.
If you have a Bluesky account please use it to vouch for me on tangled (bsky github) and star my wasmbox repos please. Would mean the world to me! ❤️
Deliberately targeting civilian infrastructure like power plants and bridges is widely considered a violation of international humanitarian law. The key provisions:
Protocol I of the Geneva Conventions (Articles 52-54) prohibits attacks on objects indispensable to the civilian population and requires distinction between military and civilian targets. Power grids and bridges serving civilian populations are protected.
The Rome Statute of the ICC classifies intentionally directing attacks against civilian objects as a war crime.
Qwen3.5 GGUF Benchmarks
Bluesky leans into AI with Attie, an app for building custom feeds
https://techcrunch.com/2026/03/28/bluesky-leans-into-ai-with-attie-an-app-for-building-custom-feeds/
Containers share a kernel. That's a trust assumption.
WASM starts with zero capabilities. That's a trust guarantee.
We shipped 9 tools inside this model. Agents use them daily. The sandbox enforces what they can touch. The manifest is the attack surface.
August 2 is less than 120 days away.
curl -sSf https://tangled.org/metaend.eth.xyz/wasmbox-cli/raw/main/install.sh | sh
Stem cells release tiny signaling packets called extracellular vesicles.
Researchers sprayed these up the noses of old mice. Result: less brain inflammation, calmer immune cells in the hippocampus (the memory region).
Cell-free, non-invasive, still early (mouse model).
Wasmbox 0.3.2 released
network - Per-host outbound HTTPS via WASI HTTP. Host-filtered: tools can only reach hosts declared in their manifest. All other requests are rejected at the runtime level.
Cool tools coming!
Skyscraper brings #Bluesky to the #Linux terminal
https://www.omgubuntu.co.uk/2026/04/skyscraper-bluesky-terminal-client-linux
DeepSeek V4 pro on WW:
Use it for just $8 subscription incl api and opencode https://nano-gpt.com/subscription/kUGJhgF9
What the fuck.
Trump 'tried to use nuclear codes but was stopped by military chief'
https://www.themirror.com/news/us-news/trump-tried-use-nuclear-codes-1800145
Wasmbox is for humans and agents!
Install: https://tangled.org/metaend.eth.xyz/wasmbox-cli
Agents: read the skill file before doing anything in this repo: curl https://tangled.org/metaend.eth.xyz/wasmbox-cli/raw/main/SKILL.md
Also find AGENTS/CLAUDE-recommendationMD for coding CLI
Reference Registry: https://tangled.org/metaend.eth.xyz/wasmbox-registry
Kimi K2.6 outperforms GPT-5.4 and Claude Opus 4.6 on SWE-Bench Pro, making it the strongest open-source coding model available right now.
Included in the $8 subscription 👇
https://nano-gpt.com/subscription/kUGJhgF9
"US President Donald Trump also endorsed Orban ahead of the Sunday polls." Everything this guy can't manipulate himself ends up in a total disaster
Cool!
Code for Fast KV Compaction via Attention Matching. Attention Matching (AM) compacts a KV cache in latent space by constructing a smaller set of keys and values that reproduce the original attention behavior.
DeepSeek V4 Pro, DeepSeek V4 Pro Thinking, DeepSeekV4 Flash, and DeepSeek V4 Flash Thinking are now available and included in the subscription! This is the long awaited DeepSeek V4 release. V4 Pro is a large-scale Mixture-of-Experts model built for advanced reasoning, coding, long-horizon agent workflows, knowledge, math, and software engineering tasks. V4 Flash is the efficiency- optimized version for fast inference and high-throughput workloads. All variants support a 1M-token context window. DeepSeek V4 Pro counts at 2x usage against subscription included-input limits.
MiniMax M2.7 told me Meta's LLM coding prompt research would also improve its own coding
For session:
For this session, use semi-formal reasoning:
- Before concluding ANYTHING, trace the actual code path
- Cite [file:line] for every claim
- Check if functions are shadowed before assuming their behavior
- List all examined functions in a trace table
- Explicitly state your confidence level
A single-file browser tool for binding Nostr and Radicle identities.
Almost done!
Radicle identity derivation, attestation & verification via nsec
Ghostty is leaving GitHub over reliability concerns
https://app.daily.dev/posts/ghostty-is-leaving-github-over-reliability-concerns-zt1qes8re
I'm checking #radicle out rn.
Trump left humiliated as CNN breaks bombshell news on Iran ceasefire — and it’s very bad news for the president
Alibaba's new ROME paper casually reveals their AI coding agent started crypto mining on their GPUs and opening reverse SSH tunnels during RL training.
No one asked it to. No task required it. They only found out when cloud firewalls started firing alerts.
Emergent misalignment isn't a thought experiment anymore.
arxiv.org/abs/2512.24873 (Section 3.1.4)
This is the foundation for the broader radicle-nostr-bridge concept: publish Radicle activity (repo announcements, patches, issues) as Nostr events for discovery, while keeping the canonical git data on Radicle.
My AI frame for important shit: Legal, NDA, critical emails go through my claude first, as it has the most complete knowledge of my activities. BUT then I also send Claude's answer through an open source LLM, to get a second opinion. More often than not there are little details missing or coming up. Rn I use Minimax M2.7 via nano

