Joined the @omarchy meetup with @dhh in Shenzhen today 🇨🇳 Loving how we’re moving in the same direction — open source, private, local-first, and agentic. Feels like we’re just getting started. Huge thanks to @brutoshi_ for hosting such a great event!

ever thought your agent was being a tool? let Jev hand it the right tool for the job instead!

Introducing Jev Router for Agentic Tools ⚡ 1. Connect your agent to 10,000+ tools and models in one MCP 2. @TypesafeAI’s Jev picks the best for your task 3. Result is 96% cheaper than Claude! Try it with zero markup now > aimlapi.com/mcp First 100 commenters will get a $10 promo code as early adopters!

Claude can crawl Reddit now! It just found TOP-5 trends in 1 min💥 We connected AI/ML API to Claude, and it analyzed 2,070,097 upvotes to find what's trending. It was 110x faster than a browser MCP, for just $0.26 totalю Crawl Reddit with your AI agent > aimlapi.com/mcp

Run Nanbeige4.3-33B locally on 32GB Mac 💻 We released our quant ladder of @nanbeige's new 33B model on Hugging Face from 35.2GB down to 13.7GB For 32GB hardware the sweet spot is our 20.3GB AD-IQ4_XS with a top-1% of 92.0% compared to BF16 GGUFs 👇 huggingface.co/AtomicChat/Na…

Kimi K3.1 Pro is the new Opus 5.5 but 3.5x cheaper Both models made Jensen Huang's robo-ducks. Opus took 3 hours, Kimi did it in 1 at a much lower price Run Kimi K3.1 Pro on aimlapi.com

Introducing Flint, @Recursive_SI’s first model: a fast local coding model built with our automated research system, the Eureka Machine. • Best-in-class fast local coding • Up to 2x faster than comparable models: 125+ tokens/sec on a MacBook Pro, 150+ on RTX Spark • Built from scratch in under 4 months Private, unmetered intelligence on your own machine. Open weights available now on @huggingface, run it locally on your favorite framework or try it for free on @opencode or @pidotdev.

Recursive Flint 4.7x faster than Nemotron 3.5 🎃 We gave Flint, Ornith and Nemotron the same prompt in Atomic Local AI Agent’s orchestration mode, where Opus plans and the local models get the task done Try Atomic Fusion -> atomicagent.io

Run UNCENSORED DeepSeek V4.1 Flash 1017 GGUFs locally 🐳 Refusals dropped from 91.4% to 0% and it still answers 86.6% of MMLU questions right vs 86.0% before uncensoring Get it here👇 huggingface.co/AtomicChat/De…

🚀 Introducing DeepInfer, our in-house inference framework, now fully open source! Powering DeepSeek-V4.1-Flash at scale with high throughput and low latency. We’re sharing our production stack to help the community lower compute costs, reduce energy use, and make AI more affordable. 🔗 github.com/deepseek-ai/DeepI… 1/4

DeepSeek's open-sourced engine hits 1.7M tok/s 🐳 @deepseek_ai released DeepInfer, which serves V4.1 Flash in production, with 325 tok/s for one chat and about 1.7M tok/s across 64 Nvidia Blackwell GPUs Run AI models locally -> atomic.chat

Welcome back. Space Bunny. 🚀 MiniMax-M3.1-Flash is coming.

MiniMax M3.1 Flash vs DeepSeek at Tesla Roadster 🚔 MiniMax M3.1 Flash re-created this viral Tesla Roadster police chase scene for $2.34, while DeepSeek V4.1 Flash's version cost $3.07 Run AI models locally or via API -> atomic.chat

Joined the @omarchy meetup with @dhh in Shenzhen today 🇨🇳 Loving how we’re moving in the same direction — open source, private, local-first, and agentic. Feels like we’re just getting started. Huge thanks to @brutoshi_ for hosting such a great event!

Love meeting so many Chinese Omarchy fans and users here at the GOSIM/Omarchy Shenzhen event! ✌️

Beam is now available across platforms, with open weights under the Apache 2.0 license - in collaboration with an ecosystem of 40+ partners. We're releasing: - Weights, tokenizer, and reference implementation - BF16, FP8 and NVFP4 weights and a DFlash2 drafter - The open source harness we built and trained Beam on: Mirror CLI Weights and model card on Hugging Face: huggingface.co/reflection