We're sharing/showcasing best of @huggingface models. Follow to stay in loop. Promoting Open-Source models.
Joined December 2025
- Tweets31.7K
- Following32
- Followers71.8K
- Likes624
Hugging Models retweeted
What if your AI agent could build and manage its own team?
That’s Mercury Bots.
A lead bot delegates work to specialist crews, each with its own model, memory, workspace, permissions and schedule.
The entire fleet runs on your machine in one background process.
Hugging Models retweeted
Anthropic spent five months building a replacement for Git, only to publish a detailed breakdown of how Git could be improved.
Congratulations. You’ve just become Git’s unpaid R&D department.
Five months of engineering. One very expensive feature request.
New on our Engineering blog: Creating in-house source control.
In May, we merged around 2,500 code changes per day. In September, we merged more than 30,000 per day. As a result, we became bottlenecked by developer infrastructure built for human (rather than agent) scale.
Here, we share how we built a new version control system—and fully migrated our monorepo from git—in five months: anthropic.com/engineering/re…
Hugging Models retweeted
Your AI agent can be powerful without treating your data like public property.
Mercury is built around local memory, explicit access scopes, and user-controlled permissions.
Mercury Cloud also does not train models on your data.
Your code. Your context. Your control.
Explore more at: mercuryagent.sh
Hugging Models retweeted
Fine-tuning usually means Stepping Outside the Normal Development Loop and into a Separate ML Workflow.
Overmind brings that closer to how developers already work: Connect the repo, Learn from real Agent behavior, Tune for the Workload, Ship, and Iterate.
That makes Specialised Models much more Approachable to Build with.
Open source: github.com/overmind-core/ove…
Worth a star ⭐
Overmind turns anyone into an AI lab.
• Builds a context graph from code and traces
• Curates training and eval datasets
• Evaluates prompts and models
• Trains smaller, specialized models
github.com/overmind-core/ove…
Hugging Models retweeted
“Open weights release end of October.” 🎉
BOOOOM WE ARE SO BACK
huggingface.co/mistralai/Mis…
Meet Mistral Large 4, aka Le Chonk.
• 1T parameters, natively multimodal. 49B active.
It is the best open weights model from US or Europe on aggregated benchmarks.
• State-of-the-art on critical workloads, including cyber defense, manufacturing and finance and it surpasses closed frontier models on visual grounding.
• Forged in Europe end-to-end and is deployable from Europe via our own Mistral Cloud infrastructure.
• Available to all via API today. Working with cybersecurity partners privately.
Open weights release end of October.
Hugging Models retweeted
Not every AI Agent task needs a Frontier Model.
Overmind turns real Agent data into Specialised Models built for the Job, with evals, fine-tuning, deployment, and model weights you own.
Open source:
github.com/overmind-core/ove…
Worth a star ⭐
Introducing Overmind. The training platform for agents.
Overmind trained models across biomedical, legal, and engineering tasks beating the frontier and achieving:
• 7x better accuracy
• 20x cheaper usage
• 28x less hallucinations
Train your own: github.com/overmind-core/ove…
Hugging Models retweeted
Public awareness:
Mercury Agent is not associated with any cryptocurrency, token, coin, or Web3 project.
Any account, token, listing, or project claiming otherwise is not linked to us. Please avoid interacting with it and report any suspicious or impersonating accounts you come across.
Official updates will only come through our verified Mercury channels.
Hugging Models retweeted
We’ve hired a new Sr. SW Developer.
Meet, mercury /code - builtin your mercury agent.
Enjoy, top-tier programming.
Hugging Models retweeted
We've spent years finding open source projects that deserve more attention and putting them in front of developers.
Some had a few stars. Some had thousands. What mattered was the work behind them.
We've watched those projects find users, contributors, and communities of their own.
Finding good open source. Helping more people discover it. Supporting the builders behind it.
To every maintainer, contributor, and developer who's been part of this: Thank You.
There's a lot more to come.
Follow @GitHubProjects.
Hugging Models retweeted
I asked Opus…
I asked Fable 5…
I asked GPT Sol…
I asked Qwen 3.6…
I asked Gemini…
I asked Grok…
I asked DeepSeek…
I asked Kimi…
I asked GLM…
I asked MiniMax…
I asked Llama…
I asked Mistral…
I asked Nova…
I asked Command R…
I asked Mercury…
I asked Ox Alpha…
I asked every model I could find…
And somehow, they all pointed to the same answer:
Hugging Models retweeted
The people behind Ox Alpha are stepping forward soon.
🚨OX Alpha is back on Mercury Cloud for free.
Some users recently hit rate limits due to unusually high demand and token usage. We’ve now reset the affected usage limits and significantly increased token capacity, so you can continue building with fewer interruptions.
Try OX Alpha again:
cloud.mercuryagent.sh
Hugging Models retweeted
We’re releasing new Qwen3.8-27B GGUFs with 10% higher accuracy.
Unsloth Dynamic V3 outperforms others by >10% on Div-300, KLD & more benchmarks.
We also release 1-bit quants that retain 77% accuracy. Run on 8GB RAM.
Blog: unsloth.ai/docs/basics/dynam…
GGUF: huggingface.co/unsloth/Qwen3…
Qwen3.8-27B can now be run locally! ✨
Run on 17GB RAM via Unsloth Dynamic GGUFs.
Qwen3.8-27B is by far the strongest model for its size. We also uploaded NVFP4 quants.
GGUF: huggingface.co/unsloth/Qwen3…
Guide: unsloth.ai/docs/models/qwen3…
Hugging Models retweeted
GLM-5.3 is now available on Mercury Cloud.
Connect your Mercury Agent and start using it instantly. No separate API key. No additional setup.
Your agent. Your model. Your rules.
cloud.mercuryagent.sh
Hugging Models retweeted
LongCat-2.0 is 60% discounted on mercury cloud.
Try now:
cloud.mercuryagent.sh
Introducing LongCat-2.0 🐱
1.6T parameters · MoE with ~48B active · 1M context
The full model behind Owl Alpha on @OpenRouter — now available.
Built for agentic coding from the ground up:
◆ LongCat Sparse Attention (LSA) — scales efficiently for 1M-context tokens
◆ Zero-Compute Experts — dynamic activation 33B–56B per token, zero wasted compute
◆ MOPD — three specialized expert groups (Agent / Reasoning / Interaction), gate-routed per task
How it stacks up:
→ Terminal-Bench 2.1: 70.8
→ SWE-bench Pro: 59.5 (GPT-5.5: 58.6)
→ SWE-bench Multilingual: 77.3
→ FORTE: 73.2 · RWSearch: 78.8 · BrowseComp: 79.9
📖 Tech Blog: longcat.chat/blog/longcat-2.…
Try it across different scenarios 🧵👇
Hugging Models retweeted
What if your AI agent could remember, collaborate, connect with your friends’ agents, and follow you across devices?
We’re dropping 10 open invites to Mercury Cloud.
Experience 300+ models, 10+ top LLMs free to use right away, Cloud ↔ Local agent connections, collaborative knowledge + memory sharing, cloud sync, Cloud Chat, Cockpit, agent management, and more.
No waitlist. First 10 are in.
Codes below ↓
MRCY-AA89953428672C3DB101
MRCY-0309ADBF47E8631229E6
MRCY-F48E6E032A42DF517896
MRCY-AED17D7D0C7851065CE0
MRCY-0A97FB4D274B5DC58328
MRCY-C81793A96F1C9DBF80A5
MRCY-A0E231401AAA45AADB75
MRCY-6BDBA5CBD12C184CC762
MRCY-2367187A1731A92837BF
MRCY-3735FF655BCA436DD58F
