The AI zombie short @PJaccetturo called "one of the best short films in years"?
The creator (Mx-Shell) open-sourced his prompts. His full method.
I translated the kit into English.
The 5-section template + 3 techniques nobody's talking about ↓
Too Powerful to Fully Release: Claude's New Model Is Here — But the Full Version Is Locked Anthropic just did something unusual: it released its most powerful AI model to the public — but kept the full, unlocked version behind a wall. On June 9, 2026, two models dropped at once: Claude Fable 5 (the "tamed" version, available to everyone) and Claude Mythos 5 (the unlocked version, restricted to trusted partners only). Same underlying brain. Two levels of access. In this video, I break down what's actually going on and what it means for developers. ⏱️ What I cover: Why Mythos was locked up in the first place (its cybersecurity capabilities) How Fable 5 differs from Mythos 5 — and the safety guardrails that route high-risk queries to Opus 4.8 The benchmarks: 80.3% on SWE-Bench Pro, 78% on ExploitBench, and why long, complex tasks are where it pulls ahead The catch: pricing ($10/$50 per million tokens — double Opus 4.8), the free window before June 23, and the mandatory 30-day data retention policy What this actually means if you're building multi-agent workflows or using Claude Code daily 📷 Key dates: Free on Pro/Max/Team/Enterprise through June 22 — then it moves to usage credits. Sources: Anthropic's official announcement + reporting from TechCrunch, CNBC, NBC News, and IT Pro (June 9, 2026). https://t.co/MoJ3sV208e
#Claude #Anthropic #AI #ClaudeFable5 #Mythos5 #AInews #LLM #ClaudeCode
Too Powerful to Fully Release: Claude's New Model Is Here — But the Full Version Is Locked Anthropic just did something unusual: it released its most powerful AI model to the public — but kept the full, unlocked version behind a wall. On June 9, 2026, two models dropped at once: Claude Fable 5 (the "tamed" version, available to everyone) and Claude Mythos 5 (the unlocked version, restricted to trusted partners only).
Same underlying brain. Two levels of access. In this video, I break down what's actually going on and what it means for developers.
⏱️ What I cover:
Why Mythos was locked up in the first place (its cybersecurity capabilities)
How Fable 5 differs from Mythos 5 — and the safety guardrails that route high-risk queries to Opus 4.8
The benchmarks: 80.3% on SWE-Bench Pro, 78% on ExploitBench, and why long, complex tasks are where it pulls ahead
The catch: pricing ($10/$50 per million tokens — double Opus 4.8), the free window before June 23, and the mandatory 30-day data retention policy
What this actually means if you're building multi-agent workflows or using Claude Code daily
📌 Key dates: Free on Pro/Max/Team/Enterprise through June 22 — then it moves to usage credits.
Sources: Anthropic's official announcement + reporting from TechCrunch, CNBC, NBC News, and IT Pro (June 9, 2026).
#Claude #Anthropic #AI #ClaudeFable5 #Mythos5 #AInews #LLM #ClaudeCodehttps://youtu.be/wZNt8jkl3QQ
Introducing Claude Fable 5: a Mythos-class model that we’ve made safe for general use.
Its capabilities exceed those of any model we’ve ever made generally available.
Why does the whole market shake when Japan just nudges rates?
The yen carry trade: borrow near-zero yen → buy higher-yielding assets worldwide. BOJ hikes → yen jumps → the trade unwinds → forced selling. Add US recession fears, and the Nikkei craters, dragging markets globally.
A 1-min plain-talk explainer 👇
https://t.co/tP96CIlX1x
#YenCarryTrade #Nikkei
Just started building in public and getting crickets? It's fine. Been at it for months and still crickets? Also fine.
I was there. For 3 years.
Joined X in Dec 2021. Lurked. Posted maybe once a quarter. 10 impressions per post, 8 of them me refreshing.
This year I actually committed. Started shipping in public — Claude Code workflows, AI agent experiments, real product decisions. Slowly, replies showed up. Then DMs.
What changed wasn't the writing. It was finding people on the same frequency.
There's no growth hack. But there is one move that compounds: Find builders who are struggling like you. Actually read their posts. Leave one useful sentence.
Not follow-for-follow. Real reads, real replies.
If you're an indie hacker still grinding — reply with what you're building. I'll go look.
I build AI agents and ship open source: → agency-orchestrator — one sentence → multi-AI collab, 211+ expert roles (TypeScript) → agency-agents-zh — 12.1k ⭐ Also #2 contributor on upstream agency-agents.
Glad you're here.
#buildinpublic #AI
@PJaccetturo@eliott__mogenet@CheHubaStyle@Stargocn@xun_mi21858 I created the full prompt system used for this *Zombie Scavenger* short film. Open source here: https://t.co/EVNaKeNlZ0 MX-Shell style prompts + consistency + cinematic controls included. Happy to share and collaborate! #AIShortFilm
@PJaccetturo@eliott__mogenet@CheHubaStyle@Stargocn@xun_mi21858 I created the full prompt system used for this *Zombie Scavenger* short film. Open source here: https://t.co/EVNaKeNlZ0 MX-Shell style prompts + consistency + cinematic controls included. Happy to share and collaborate! #AIShortFilm