✨ QQL Minting Reopened! Last 50 passes are free - take a look: x.com/Senat_Hamburg/status/2… @AlexEstay780265 @mindlab @hankmarsalis x.com/i/broadcasts/xz6JUiyjk…

Don't forget! Our next MindLab is tomorrow, Monday, June 15, with five different streaming times. Join us for a live mental performance coaching session on the subject of “adaptability.” For youth athletes — individuals and teams. Your first month is free at bit.ly/ACGmindlab.

Mind Lab has open-sourced our first model, V1-Preview 🎉 See also: macaron.im/mindlab/research/… #mindlab #macaronai #agents #lora

Liz Kakooza’s MindLab is reimagining wellbeing leoafricainstitute.org/news/… Africa doesn't have a mental health awareness problem. It has a mental health infrastructure problem. 1.4 mental health workers per 100,000 people. That's the continental average. The global average is 9.0. Between 2000 and 2015, Africa's population grew by 49% — but years lost to disability from mental and substance use disorders grew by 52%. The crisis is outpacing the systems meant to contain it. And yet the conversation, when it happens at all, tends to stop at stigma. Stigma is real. But stigma is not why governments fail to fund mental health services. Stigma is not why nearly half of African countries lack functional standalone mental health policies. Stigma is not why untreated depression alone is estimated to cost Uganda hundreds of millions of dollars annually in lost productivity. That is a governance failure. And framing it as anything less lets the right people off the hook. Liz Kakooza @lizkk has been making this argument for over a decade: in policy spaces, in workplaces, at the World Economic Forum, and in the communities her organisation #MindLab is designed to serve. Her position is unambiguous: wellbeing is not a wellness benefit. It is a business imperative, a leadership responsibility, and a justice issue. When wellbeing is treated as a perk, it is the first line cut when budgets tighten. When it is understood as a structural condition for human performance and dignity, it becomes non-negotiable. Africa has the knowledge, the community infrastructure, and the leaders to close this gap. What it needs is the political will and institutional courage to act on what it already knows. #AfricaChampionsNetwork #DistinguishedFellow

Here's an assumption worth interrogating: that the solutions to Africa's development challenges sit somewhere outside Africa, waiting to be imported. @lizkk Kakooza doesn't accept that premise, especially not in the field of mental health and wellbeing. What she argues, and what her decade of work demonstrates, is that the answers already exist within African communities. The challenge is not importation. It is excavation: building the policy frameworks, the institutions, and the leadership required to surface those answers and make them work at scale. That distinction matters more than it might first appear. Imported frameworks carry hidden assumptions about what a household looks like, what a workplace relationship means, how trust is built, what counts as healing. When those assumptions don't match the lived reality of the communities they're meant to serve, the frameworks fail quietly, and the gap widens. Community-led approaches built from within don't carry those assumptions. They carry something else: legitimacy. Ownership. Durability. This is what MindLab is doing in Uganda. It's what BasicNeeds Kenya is doing. What the Mentally Aware Nigeria Initiative is doing. What a growing generation of African leaders is beginning to demonstrate across the continent: that the most durable solutions to African challenges will be African in origin, not just in geography. The work is harder that way. It asks more of its practitioners. But it tends to last. 💡 #AfricaChampionsNetwork #DistinguishedFellow

Directly to the left of Eris' It has a small version of the Mindlab structure, a screen with Warmind code and the head of the Exo body from Season of the Seraph

Τα είχε πει όλα #mindlab youtube.com/shorts/By4_6VZOp… via @YouTube

Macaron-V1-Preview-749B 👀 a Mixture-of-LoRA personal agent model from MindLab ✨ 744B base 5 specialist LoRAs ✨ Generative UI as a core skill ✨ Personal agent focused ✨ 202K context ✨ MIT license

🚨 Stop wasting money on bigger models. Here are the best open-source AI models for your hardware this week 👇 If you're building AI agents, coding tools, or local AI setups, save this. 8–12GB VRAM → [LiquidAI LFM2](huggingface.co/LiquidAI?utm_…) • Insanely fast • Tiny footprint • Surprisingly capable for its size Perfect for laptops and entry-level GPUs. 16–32GB VRAM → [Google Gemma 12B](huggingface.co/google?utm_so…) • Performance rivals models nearly 2x larger • Strong reasoning • Excellent efficiency Also worth watching: → [JetBrains AI Models](huggingface.co/JetBrains?utm…) Currently among the best performers on LiveCodeBench. 32–96GB VRAM → [Nex-N2-Mini](huggingface.co/nex-agi/Nex-N…) • Qwen-35B post-trained model • Strong "GPT-style" behavior • Emerging class leader for reasoning → [Qwopus by Jackrong](huggingface.co/Jackrong?utm_…) • #1 Q4 quantized Qwen variant on several agent coding benchmarks • Strong coding and tool-use capabilities 192GB VRAM → [Step-3.7-Flash](huggingface.co/stepfun-ai?ut…) • Fast inference • Vision capable • Newer knowledge cutoff • Hard to beat for the hardware requirement 384GB VRAM → [Nex-N2-Pro](huggingface.co/nex-agi/Nex-N…) • Massive Qwen-3.5 post-train • Reported state-of-the-art performance on DeepSWE-style evaluations 768GB VRAM → [MindLab GLM Post-Train Models](huggingface.co/mindlab-resea…) • Strong benchmark results • One of the most promising large-scale open model releases right now The pace of open-source AI is getting ridiculous. A few months ago, many of these performance levels required much larger hardware budgets. Today, smaller models are catching up fast. Which model are you running in production right now? Follow @smratitiwa86867 for daily AI, Agents, MCP, Open Source AI & Automation insights. 🔥

Hmm I wonder if we quantize down mindlab-research/Macaron-V1-Preview-749B if it would still perform well on a M3 Ultra 512 GB 🤞