Axel Dittmann
@DittmannAxel
Followers
15K
Following
7K
Media
497
Statuses
5K
🌅 Day:AI/Cloud Architect @Microsoft | 🌙 Night: CEO Dittmann-Digital GmbH 🔧 Building: GenAI Solutions -all statements are private and my own opinion
Duesseldorf
Joined July 2009
Found a prompt somewhere on X and modified it with my two dogs - #VEO3 is insane good!! Dogs are look alikes :-) :-) - and the best: available in Europe now!! #Innovation #GenAI
1
0
1
I’m just reading that Ryzen AI 395 has to be 30% slower than DGX Spark in LLM inferencing… and only 96GB GPU RAM… good I haven’t RTFM upfront, so I made the AMD faster with 128GB unified RAM 🫡 Z2 mini G1a can run Qwen3 Coder 30B BF16 at 26.8 tok/sec in ~60GB GPU RAM
0
1
2
Hocus Pocus and 128GB of GPU VRAM appeared on my AMD Ryzen AI 395. Qwen3 Coder 30B at BF-16 with 1Million token context window running at 24.4 tokens/sec Codebase semantic-search embedding model Guardrail so it’s all legit Total GPU footprint 115GB, still 13 GB left for Linux
0
1
2
🧠🤖 When agents start creating agents… during runtime. Forget static workflows. With the Microsoft Agent Framework, an LLM can design and launch entire processes dynamically — no predefined flow, no manual setup. Not automation. Autonomy. #GenAI #AI #Agents #Microsoft
0
0
1
Browsers are fucking hard yall, negative tweets are easy. I've worked on a novel chromium based browser for 2 years (it never shipped) and I can tell you, with confidence, they didn't just "fork chromium and slapped chatGPT around it". Shipping it on one platform only makes
@SimonHoiberg They took Chromium. Forked it. Wrapped Chatgpt around it. Fine-tuned system prompt. And ONLY shipped in MacOS! This is such a low effort stab at what is inarguably one of most complex and sophisticated market - A Browser. The 500 billion dollar company, ladies and gentlemen.
20
9
150
I love Grok :-) Grok Imagine prompt: the black dog asks: hey where are we going, the brown dog asks I wanna know, too - the man responds: home - the walk through the jungle and birds are tschirping
0
0
0
awesome - can't wait to see the results
Say hello to my little friends! I just unboxed this trio of HP Z2 G1a! Three is always better than one! 3x AMD Ryzen AI Max+ Pro 395 384GB RAM 24TB of RAID storage ROCm 7.0.2 tonight llama cpp, vLLM and Aibrix
0
0
2
10 am livestream today to launch a new product I'm quite excited about!
1K
808
11K
DeepSeek-OCR is out 🔥 https://t.co/MIqKMWsPwO ✨High-accuracy OCR - MIT license ✨Fast GPU inference (FlashAttention 2, BF16) ✨Docs > Markdown ✨Works with transformers
huggingface.co
24
163
1K
🔥 Just ran DeepSeek-OCR locally — full image analysis + text extraction in 11.1 seconds. That’s not faster OCR. That’s document AI in real time. Next up: Mistral OCR — it reads text, tables, and equations in one pass. Run it hybrid on Azure to scale out on demand while keeping
0
0
1
Friday evening. KAT-Dev-72B-Exp is spinning in Aibrix K8s. The GPUs in the Z8 are fired up. It's a LAN party for one. After 6 months on a diet of MoEs, I'd forgotten the main-course feeling of a dense 72B model.
1
1
6
👀 The ultra-talented team at @BlackForestLabs (behind great open models like FLUX 1.1 and Stable Diffusion) is hiring a growth marketer, and telling it like it is on the job description! More of this mindset, please (at every company). 🔗 Link below, to appease the algorithm:
2
1
12
@KwaiAICoder is there an example of how to inference it in vLLM or SGLang with tool use?
1
1
2
Factory has opened the Droid’s coding agent model cage to support open models local AI. Let’s give it a spin. On LAN with GLM-4.6 served via Aibrix vLLM on my Z8 GPU workstation Offline on my MacBook Air with Qwen3-Coder-30B-3AB quantized to the size of a Taylor-Swift ringtone
0
1
3
I'm coding with Factory Droid, offline on my MacBook Air. Qwen3 Coder Flash as LLM, no Wi-Fi, no cloud, no AI API bill; just a fantless 1.24 kg slab of aluminum with a GPU. It generates usable code and tool calls at a cool ~20 tokens/sec.
0
1
4
Factory Droid using IBM Granite 4.0 H Small on llama.cpp, tools, and internet via MCP is chugging the NVIDIA dynamo monorepo. 12 mins and 0$ for AI
1
2
6
.@Microsoft debuts its Agent Framework – a unified, open-source SDK merging AutoGen and Semantic Kernel for enterprise multi-agent AI systems. Backed by Azure AI Foundry, it simplifies orchestration, observability, and plays well with any API. - Private preview of multi-agent
3
4
34
Introducing AgentKit—build, deploy, and optimize agentic workflows. 💬 ChatKit: Embeddable, customizable chat UI 👷 Agent Builder: WYSIWYG workflow creator 🛤️ Guardrails: Safety screening for inputs/outputs ⚖️ Evals: Datasets, trace grading, auto-prompt optimization
277
2K
11K