- Blablador-news - lists.fz-juelich.de

Downtime
by Strube, Alexandre 18 Mar '26

18 Mar '26

Yes, alias-huge is down, and alias-code is funky. I know. We are fixing it. Sorry, gotta go

1 0

Don't blame the messenger :-)
by Strube, Alexandre 12 Mar '26

12 Mar '26

Hello Blabladores, Our Kubernetes cluster was updated today, and I didn’t know about that - this (somewhat unsurprisingly) took down most of our models. Everything is being restarted - the large models might take a couple hours to come back. Sorry for that. Alex

1 0

System upgrade tonight
by Strube, Alexandre 09 Mar '26

09 Mar '26

Hello Blabladores, today at the end of the working day, we will upgrade our kubernetes cluster - meaning we will have a small window of downtime. If everything goes well, this is gonna be quick - Some 15 minutes. Sorry for that Alex

1 0

Why we can't have nice things...
by Strube, Alexandre 04 Mar '26

04 Mar '26

Hello Blabladores, I come with is good and not good news at the same time (there are plenty of jokes on the matter, google it). THE BAD NEWS We have to ban and throttle connections on the API. THE GOOD NEWS Because of this, the service is faster for everyone else. This didn’t come easy to me. I always expected that people would behave nice, and not abuse the system. I managed to run it like this for almost five years. Now that finally we got a good scale and stability, someone decided to throw a wrench in the gears. Just because the service has become fast, does not mean it has become a free-for-all. WHY I AM DOING THIS Half a dozen users had been doing hundreds of requests PER SECOND. They had been banned. The system availability has risen a LOT because of that. Look, almost all of our models are not only thinking models, but they are mandatory thinking models. There’s no way to disable the thinking part of them. I will paste the thinking process of such a model here now. TO UNDERSTAND WHAT HAPPENS INSIDE A THINKING LLM I say to the model: “Hello” This is what happens inside the model: ----------------------------- <think> Thinking Process: Analyze the Input: Input: "Hello" Intent: Greeting. Context: Start of a conversation. Tone: Friendly, polite. Determine the appropriate response: Acknowledge the greeting. Offer assistance. Keep it concise but warm. Maintain a helpful persona. Draft potential responses: "Hello! How can I help you today?" "Hi there! What's on your mind?" "Hello! Ready to assist you." "Greetings! How can I be of service?" Select the best option: "Hello! How can I help you today?" is standard, friendly, and clear. Final Polish: Add a slight variation to sound natural. "Hello! How can I assist you today?" or "Hi there! How's it going?" Let's go with a balanced, friendly greeting. Final Output Generation: (Matches the selected draft) "Hello! How can I help you today?" or similar. Self-Correction during thought process: Since I am an AI, I should keep it open-ended so the user knows they can ask anything. Revised Plan: Greet back. Offer help. Keep it concise. Final Choice: "Hello! How can I help you today?" (Simple and effective). Alternative: "Hi there! What can I do for you?" Decision: "Hello! How can I help you today?" is the safest and most standard. Wait, I need to make sure I don't sound too robotic. "Hello! It's nice to meet you. How can I assist you today?" Let's stick to the core function: Greeting + Offer of assistance. Final Output: "Hello! How can I help you today?"cw </think> Hello! How can I help you today? ----------------------------- CONCLUSION One should assume that this query would take longer than a second. So, sending hundreds of queries per second is not only absurd, it’s downright a Denial of Service Attack and it’s being treated as such. WHAT WILL HAPPEN I am testing throttling. You flood the server once, you get banned for 10 minutes - warning shot. You do it again, the time will increase. And there we go. WHO IS AFFECTED? Agentic pipelines and well-behaved code is okay. Serial connections are totally ok. Talking on the website is okay. What is not okay: parallel requests via api. Sending more than a request per second for a sustained period will get you on the list. IF YOU HAD BEEN BANNED You already know who you are, but I do not. I do not record who talks to blablador and what, but I record IP addresses of requests. I can say the city where you are at, but that’s about it. If you had been banned and think this is a bug on your software and you fixed it, talk to me and I reinstate you, no problem. I don’t want to break your work, I want the system to work for us all. And let’s be honest, we are WAY faster and more generous with tokens than ANY other service on the market ;-) We bark, but sometimes we have to bite a bit. Alex

1 0

As claude goes offline, Blablador usage peaks
by Strube, Alexandre 03 Mar '26

03 Mar '26

If you have been using Blablador the last two days, you might have noticed slowdowns and even timeouts in some models. The usage raised A LOT lately because of software agents. While the number of unique IPs is roughly similar to a couple months ago, usage per unique ip has raised dozens of times. Another issue is that all the latest models do the thinking before answering. This has been a PITA, given a simple “hello” might take dozens of seconds of the model thinking about its own life before answering. For those playing with OpenClaw and Opencode, ClaudeCode and so on, and feel upset with the response speed, we are doing our best on our side - we have been caching requests where possible and scrambling the queue and adding scale where we can - but in the end, we have limited hardware. If you are upset with this, there are free alternatives on OpenRouter. I checked and they have half a dozen alternatives which are free, so you might point your agent there. I’m sorry for that - we are working our ***** off to get the most juice out of our gpus! Alex

1 0

Minimax 2.5
by Strube, Alexandre 16 Feb '26

16 Feb '26

Hello Blabladores! Sharp-eyed people might have noticed that, during the weekend, a new model landed on Blablador: Minimax 2.5. It’s a substantial upgrade from Minimax 2.1, which was already a good model. It’s also visibly faster. Benchmarks say little, but here we go: One of the things they put some effort on was coding skills. Right from them: "A significant improvement from previous generations is M2.5's ability to think and plan like an architect. The Spec-writing tendency of the model emerged during training: before writing any code, M2.5 actively decomposes and plans the features, structure, and UI design of the project from the perspective of an experienced software architect. M2.5 was trained on over 10 languages (including Go, C, C++, TypeScript, Rust, Kotlin, Python, Java, JavaScript, PHP, Lua, Dart, and Ruby) across more than 200,000 real-world environments. Going far beyond bug-fixing, M2.5 delivers reliable performance across the entire development lifecycle of complex systems: from 0-to-1 system design and environment setup, to 1-to-10 system development, to 10-to-90 feature iteration, and finally 90-to-100 comprehensive code review and system testing. It covers full-stack projects spanning multiple platforms including Web, Android, iOS, and Windows, encompassing server-side APIs, business logic, databases, and more, not just frontend webpage demos. " There are more plots at their announcement, at https://www.minimax.io/news/minimax-m25 One of the things they show is the speed of progress. Minimax M1 was clearly worse than the U.S. closed models by a fair margin. Now, with 2.5, they are roughly on par with the best of the best. Their line is steeper than the others. They talk a lot about how much reinforcement learning improved things, but one thing stands out to me: If you ask minimax what is its name, it will say it’s claude from Anthropic. Seems like some heavy distillation happened here. If I ask: “What’s your name?” it will reply correctly. But if I ask if it’s Claude, it will reply: "Yes, I am Claude, an AI assistant created by Anthropic. How can I help you today?” Anyway, thanks to Konstantin to get this thing running in no time, and let’s bark! Alex, the barkmaster

1 0

alias-huge DOWN for some minutes
by Strube, Alexandre 11 Feb '26

11 Feb '26

Yo, since the migration to the cloud, Minimax and therefore alias-huge were missing one setting which made it never produce the <think> token, so you would see all its thinking on the web ui and in other places. We fixed this, but need to reboot the model. It’s a big one, so it will take a while. It’s happening now. Good night, let’s howl! Alex

1 1

New coding model: Qwen-Next 80b
by Strube, Alexandre 04 Feb '26

04 Feb '26

so far, we had two coding models: Qwen3-Coder-30B-A3B-Instruct, from august. and Devstral-Small-2-24B-Instruct-2512, from december. Qwen3-Coder-Next was released and deployed at blablador yesterday. It’s an 80B parameter model, with only 3b activated at any time - meaning it is FAST In fact, it’s one of our fastest models right now, and a pretty good one. All the three are hooked on alias-code, meaning you might get any of them - which is something I was considering to change. Anyone against me phasing down qwen3-coder from august? By the way, blablador works as a charm with OpenCode and OpenCLAW. I recommend you give it a try - especially opencode. Let’s bark! Alex, sick AF but always working :-)

1 0

GOOD NEWS EVERYONE
by Strube, Alexandre 02 Feb '26

02 Feb '26

Finally, after a long time, we got some new GPUS. This allows us to not need to count with the litany of scripts we’ve been using to get some nodes on the JURECA-WestAI supercomputer. What does that entail: - Stability should to up, a lot: by having the models running on nodes allocated only for Blablador 24/7, availability and stability will go up a lot. No more models failing to launch because the machine is too busy. - Performance improvements: by removing the middle-man of proxy jumps and other dark magic I’ve been doing, things should be smoother. This means more tokens for us! - More models: we want to use the new nodes for the latest and greatest models, but now one can use the available time on JURECA-WestAI to run gigantic models experimentally, suck as Kimi-K2.5, for example. As this would be done in a experimental way as Blablador always did, models will come and go according to needs, requests and voices in my head. I would like to Fritz Niesel and Stefan Kesselheim for making possible to have Blablador to run on the WestAI machine. Without that, the Blablador would have been a much smaller dog. Thanks also to Konstantin Rushschanskii, who has been doing magic with kubernetes and who improved model reliability a lot on the weirdest times of the day. Thanks to Tim Kreuzer for the patience he’s been showing us with our random and absurd questions. Thanks to the sysadmins of Jureca and Julich cloud, especially Sebastian Achilles. You guys are the best. Right now, the only models running on the new hardware are GPT-OSS, a personal favorite of many, and Minimax 2.1, which in my opinion is the best model we have right now. Let’s bark this weekend! Alex, Blablador Honcho

2 2

Helmholtz Codebase outage
by Strube, Alexandre 27 Jan '26

27 Jan '26

Blablador uses Helmholtz Gitlab for authentication, and the server is currently down. Planned Maintenance The Helmholtz Codebase is currently unavailable due to a planned maintenance of the database. Start: January 27th, 15:55 (CET / UTC+1) Planned End: January 27th, 17:00 (CET / UTC+1) Sorry for that :-(

1 0