Stephen McConnachie

@NousResearch @ogichain @FFmpeg it’s always @ffmpeg underneath

🥇 1st Place: @ogichain - Media Tool A video/audio editing toolset that registers as a native Hermes skill. Send a file in chat, describe what you want in plain language, get the edited result back. Trim, compress, convert formats, burn subtitles, generate GIFs, make Telegram stickers, normalize audio — all through conversation. Chain multiple operations in one message: "trim 0:30 1:00 and speed 1.5x and compress." All processing runs locally through ffmpeg. Works across Telegram, Discord, WhatsApp, and CLI. This won because it solves a real, everyday problem cleanly. No more cycling between CapCut, random converter sites, and separate editors for each format. The tool registration into Hermes is well-designed (the agent routes commands automatically) and the whole system is modular enough that adding new media operations is straightforward. Practical tool use at its best. x.com/ogichain/statu…

The Hermes Agent Hackathon has ended! Finalists were selected by Nous staff out of 187 submissions on creativity, usefulness, and presentation. We were blown away by the quality and range of what people built using Hermes - thank you to all who participated. Winners below ↓

I'm excited to open source Chandra OCR 2! - 85.9% (sota) on olmocr bench - 90+ language support w/benchmarks - 4B model (down from 9B) - Full layout information - Extracts + captions images and diagrams - Strong handwriting, math, form, table support

@VikParuchuri This looks amazing! Can’t wait to try it. Would love to explore the licence terms for a non-profit charity that gets government grant and some income from our activity

We've been building an internal Claude Code plugin system at Intercom with 13 plugins, 100+ skills, and hooks that turn Claude into a full-stack engineering platform. Lots done, more to do. Here's a thread of some highlights.

@JuanG294 @travispatton @mitchellh Thanks! Will try this later in the weekend. Summon is Mac only right? Thinking about a Linux v

@travispatton @mcnatch @mitchellh ok @travispatton @mcnatch it took me more than 10 min, but there you have it. You can set up your own powerline with #Starship within your #Summon setup. It's ready to use in v0.6.1. Let me know what you think!

To celebrate that #GhosttyOrg and @mitchellh launched 1.3.1 with the fix for that annoying text grab bug, I'm launching summon, the CLI that launches your full dev workspace in Ghostty with one command 🥳 Summon your Ghostty workspace! Native splits. No tmux. No new terminal to learn. github.com/juan294/summon

<Adam Curtis voiceover> Here, in a room in Westminster, MPs practice their steps. Forward. Back. Forward. Back. Meanwhile, at the exact same moment in Tehran, a small group of newly promoted regime leaders meets to plan their own move forward...

Adam Curtis watching this knowing he can splice it in with a video of Iran and Israel tearing lumps out of eachother whilst Aphex Twin plays over it

@mervenoyann Thank you! Been waiting for this one

two more chapters on vision language models book is out! > document AI chapter (by yours truly) shows old models, new VLM approaches, retrieval and more! > video language models chapter shows video understanding, know-hows, approaches and more! sneak peek below

@vanstriendaniel Think I found it - #private-repositories" target="_blank" rel="nofollow noopener">huggingface.co/docs/hub/en/re… - thanks again

@vanstriendaniel This is so useful, thank you! Is there an obvious place where the HF data privacy / model training policies are described?

Re-OCR'd the complete 1771 Encyclopaedia Britannica (2,724 pages) with a single command on @huggingface Jobs. - 0.9B model (GLM-OCR) ~$0.002/page ~$5 total on an L4 GPU Before (old Tesseract ocr) → After:

@daemon404 So sorry!

Reviving this account to say: Almost everyone at Vimeo was laid off yesterday, including the entire video team. If you're looking for talented engineers, there are a few on the market. Sucks to see something I built killed by private equity in a technology company skin suit.

If you’re considering automating some Archivematica workflows in 2026 then here are some utility scripts I built for BFI colleagues, that might be of assistance! github.com/bfidatadigipre… Happy New Year! #digipres #Archivematica #AtoM #borndigital #python

I always like to look at Woody Guthrie's New Year's Resolutions from 1942. Pretty similar to what mine are each year.

@digitensions @MediaArea_Net @FFmpeg @BFI @MatroskaOrg Hope you have a lovely break while these servers blast through their work

Lovely to be heading out on my seasonal holidays knowing open-source stalwarts #RAWcooked (@MediaArea_Net) and @FFmpeg will continue preserving @BFI film scans to #FFV1 MKV @MatroskaOrg while resting! Thanks to all the open-source contributors!

Today marks 25 years since the first commit to FFmpeg by Fabrice Bellard FFmpeg was made to play DVDs, DivX and other video files for free, and continues to be developed by enthusiasts FFmpeg changed the world, powering all online video Happy 25th Birthday FFmpeg! 🎉🎁🎂

Last year Molmo set SOTA on image benchmarks + pioneered image pointing. Millions of downloads later, Molmo 2 brings Molmo’s grounded multimodal capabilities to video 🎥—and leads many open models on challenging industry video benchmarks. 🧵

NVIDIA releases Nemotron 3 Nano, a new 30B hybrid reasoning model! 🔥 Nemotron 3 has a 1M context window and the best in class performance for SWE-Bench, reasoning and chat. Run the MoE model locally with 24GB RAM. Guide: docs.unsloth.ai/models/nemotro… GGUF: huggingface.co/unsloth/Nemotr…

@ByteDanceOSS huggingface.co/ByteDance/Dolp…