Creative & Visual Media
Javi Lopez highlighted the rapid success of a fully AI-generated movie trailer by The Dor Brothers, reaching 3 million views in under three hours, emphasizing the creator’s superior editing skills and taste in generative video production.
3M views in less than 3 hours.
I’ve been talking with Yonatan, its creator, and basically he told me: THE MAGIC IS IN THE EDITING.
The Dor Brothers definitely have TASTE.
I won’t say that Hollywood should worry 😉pic.twitter.com/jUSGyzPI3A
— Javi Lopez ⛩️ (@javilopen) February 16, 2026
Me counting every 15 seconds to see each sequence you PERFECTLY generated. AMAZING! pic.twitter.com/2IZfxwaB0s
— Javi Lopez ⛩️ (@javilopen) February 16, 2026
He demonstrated sophisticated generative video editing techniques in response to discussions on the need for AI-driven editing tools like “Pulp Fiction cuts.”
SURE! Marchando:pic.twitter.com/xEWMZuwAXe
— Javi Lopez ⛩️ (@javilopen) February 16, 2026
Fal announced that NVIDIA’s PersonaPlex, a full-duplex model enabling simultaneous listening and speaking with diverse voices, interruptions handling, and consistent personas defined via text prompts, is now live on their generative media cloud.
🚨 NVIDIA PersonaPlex is now live on fal!
👥 A full duplex model that listens and speaks at the same time
✨ Diverse range of voices and define any role through text prompts
🎯 Handles interruptions, backchannels, and authentic conversational rhythm
⚡️produces natural,… pic.twitter.com/Mi9BDWFzeC— fal (@fal) February 16, 2026
User interfaces are increasingly likely a thing of the past? https://t.co/aiqEW5y6vM
— @levelsio (@levelsio) February 16, 2026
A.I.Warper highlighted a video demonstrating a polished hybrid AI-real animation style akin to Roger Rabbit, questioning if such approaches could gain traction in content creation.
Dope.
This got me wondering now if I’d actually enjoy watching a polished hybrid style like this.
Feels like no one has done it since Roger Rabbit
— A.I.Warper (@AIWarper) February 16, 2026
Software Development
OpenCode, the open source coding agent, welcomed James Long on his first day, sharing a video of him starting at the company amid buzz around AI-assisted coding tools.
Welcome James! https://t.co/RAGJZV4bAd
— OpenCode (@opencode) February 16, 2026
Levelsio detailed spending on Claude Code and OpenClaw subscriptions for development workflows.
I spend $100/mo for coding and then a second sub for my OpenClaw $100/mo
— @levelsio (@levelsio) February 16, 2026
Machina praised OpenAI for competing on the best coding model alongside Anthropic.
OpenAI is so back
> competing with Anthropic for the best coding model
> now building the future of personal agents with openclaw
> investing in open source
> best deep research and by fargg
— Machina (@EXM7777) February 16, 2026
Peter Omallet updated Desloppify, an agent toolset for refining “slop code,” now with C# support and security issue detection, performing well even with Codex.
Thanks to @tobitege for adding C# support! https://t.co/imFokH7Krq pic.twitter.com/23Jp1iwLSM
— POM (@peteromallet) February 16, 2026
One thing I added over the weekend to Desloppify is the ability to find all kinds of security issues.
I also hadn't tested it with Codex but it surpisingly seems to work better with it than Claude Code! https://t.co/imFokH7Krq pic.twitter.com/iCLHXn5QHC
— POM (@peteromallet) February 16, 2026
Automation & Orchestration
Omar Sar discussed CogRouter, a framework dynamically adjusting reasoning depth for LLM agents based on cognitive levels, achieving 82.3% success on benchmarks with a 7B model using 62% fewer tokens than GPT-4o.
Interesting new work on adaptive reasoning depth for LLM agents.
Not every agent step requires the same level of thinking. Some steps need strategic planning. Others are routine execution.
This research introduces CogRouter, a framework inspired by ACT-R cognitive theory that… pic.twitter.com/7Lh84Ro9OO
— elvis (@omarsar0) February 16, 2026
He also covered a benchmark on agents generating procedural knowledge via “Skills,” finding curated skills boost performance significantly (up to +51.9pp in healthcare) but self-generated ones offer no benefit on average.
Nice paper studying whether agents can generate their own procedural knowledge.
This is very important to build more reliable self-improving agents.
The new benchmark evaluates how well Skills help LLM agents across 86 tasks and 11 domains.
Finding over 7,300 agent… pic.twitter.com/YUVfu32Msk
— elvis (@omarsar0) February 16, 2026
Levelsio questioned if user interfaces are becoming obsolete, quoting frustrations with OpenClaw’s lack of UI and noting its design to eliminate them, sparking replies affirming agentic shifts away from traditional interfaces.
User interfaces are increasingly likely a thing of the past? https://t.co/aiqEW5y6vM
— @levelsio (@levelsio) February 16, 2026
Riley Brown detailed his strategy for deploying 9-12 narrow OpenClaw agents to manage growth tasks at Vibecode, with agents sharing a Notion notebook for coordination—one running on a Mac Mini and others in the cloud. He plans to document the process via videos on YouTube and @vibecode_edu, with agents posting from @vibeclaw.
Current @openclaw plan for the growth team at our startup @vibecodeapp.
Build 9-12 narrow agents that share a notebook (I'm using Notion).
One will have access to a mac mini, the others will run in the cloud.
Will be basically documenting this whole thing.
Making videos… pic.twitter.com/8rpPg5cOv9
— Riley Brown (@rileybrown) February 16, 2026
Brown also noted observing every team member at Vibecode HQ simultaneously prompting multiple agents.
Such a strange moment when i stood up at the office and I obseved every single person at @vibecodeapp HQ was prompting multiple agents.
— Riley Brown (@rileybrown) February 16, 2026
Machina (EXM7777) highlighted OpenAI’s push into personal agents via OpenClaw.
OpenAI is so back
> competing with Anthropic for the best coding model
> now building the future of personal agents with openclaw
> investing in open source
> best deep research and by fargg
— Machina (@EXM7777) February 16, 2026
Strategy & Ecosystem
Guillermo Rauch announced the close of applications for the Vercel AI Accelerator program, which received 2,275 submissions from teams eager to learn, build, and ship AI projects with over $6M in credits from partners like Vercel, v0, and AWS.
We've received 2,275 applications. Look forward to reviewing them with the team. Today is your last chance to apply! https://t.co/74xHU5VDUH
— Guillermo Rauch (@rauchg) February 16, 2026
Dan Shipper shared a video explaining the vision for @every, positioned as the essential subscription for staying ahead in AI developments.
we made a little video about why we’re building @every: pic.twitter.com/IemNgKJB56
— Dan Shipper 📧 (@danshipper) February 16, 2026
Levelsio reflected on unpredictable AI ecosystem flips, from Claude Code dominance and Anthropic favoritism to OpenClaw’s surge, an Anthropic DMCA takedown backfiring by pushing creator @steipete toward OpenAI’s Codex, reshaping narratives rapidly.
I keep realizing things flip so fast in AI you really can't predict who will win or lose
Claude Code was leading for every dev, Anthropic were the good guys, OpenAI were becoming the bad guys buying up all the RAM
I even switched from ChatGPT considering how ugly I felt the…
— @levelsio (@levelsio) February 16, 2026
OpenRouter announced MiniMax_AI’s M2.5 as the platform’s most popular model.
.@MiniMax_AI M2.5 is now the most popular model on OpenRouter pic.twitter.com/J0GivyYdjW
— OpenRouter (@OpenRouterAI) February 16, 2026
Ethan Mollick reflected that AI labs’ 2023 hype has materialized in 2026, lending credibility to their 2028 predictions.
The crazy part is that the AI Labs have generally been right. Like, the stuff they hyped in 2023 turned out to be real and working today. That doesn't mean that the stuff they are predicting for 2028 will also be real, but it is probably worth noting those predictions & watching.
— Ethan Mollick (@emollick) February 16, 2026
Machina celebrated OpenAI’s momentum in coding models, agents, open source investments, and deep research.
OpenAI is so back
> competing with Anthropic for the best coding model
> now building the future of personal agents with openclaw
> investing in open source
> best deep research and by fargg
— Machina (@EXM7777) February 16, 2026
Machina also promoted a life-changing AI podcast.
this podcast will change your life pic.twitter.com/aoArm8hYeL
— Machina (@EXM7777) February 16, 2026