Sabitlenmiş Tweet
Today we’re featured in @LaVanguardia talking about AI, data sovereignty, and why the EU is competing in one of the most important races of our time and how at @helmcode we’re working to help meet that challenge.
English
Helmcode
39 posts
Helmcode
@helmcode
🚀 AI model inference for businesses (Helmcode) and the community (https://t.co/ciWFdCqcvm)
Tenerife, España Katılım Mart 2026
19 Takip Edilen525 Takipçiler
👀☕
First Café Helmcode?
Borja Perez@borjaperfra
Oye, ya se ha hablado por aquí, pero estamos pensando en montar una primera edición de ☕ Café @helmcode. La idea es juntarnos, de manera informal, usuarios de IA abierta y gente interesada en construir software con IA en un espacio tranquilo y sin agenda. Tipo los Café Cursor pero más centrado en Open Source :) ¿Encaja? Charlar de lo que estás montando, apoyarte en otros builders, preguntar cosas... No tengo nada montado. Quiero testar interés. Si veo que interesa, pensamos dónde hacerlo 👀👀 @egraciani nos echáis un cable?
Nederlands
📸 We've created a Guide about How to prompting images like a professional.
Same idea. Same model. Different prompt.
The difference between an image that screams "AI" and one that doesn't isn't the model you use.
80% of the result is decided before you hit Generate. Here's how to use that 80%.
The full guide includes a copy-paste meta-prompt that turns any idea (or reference image) into a ready-to-use prompt, plus fixes for weird hands, invented text and "AI face".
It's useful and simple :) Ready to improve your creative skills.
Works on Helmcode and on any image model: helmcode.com/posts/guia-pro…
English
Helmcode retweetledi
1. Soberanía de modelos
2. Gasto controlado
3. Cumplir con Europa (está jode porque es un error de Europa en su planteamiento de IA. La privacidad está bien, aunque eso les da igual. No invertir en infraestructura está mal.) 🤖
@barckcode 👏👏👏👏👏👏
Helmcode@helmcode
Want to understand what we do at Helmcode, how inference works, and where open models are heading? @Saulgmz_ covered all of it in one video-blog, including an interview with our CEO & Founder @barckode. He also took a 9-hour bus trip to join our first in-person event, just to make it. He's the GOAT 🐐 The community we're building at nan.builders is insane. This is the proof. Saul deserves to be our first NaN Ambassador :) Watch the video, like, and subscribe to his channel! (Vídeo it's in spanish) youtube.com/watch?v=F4bFmY…
Español
All of this updates weekly at helmcode.com/opendata.
Numbers you can check beat claims you have to trust :)
English
WOW! We just crossed 500,000M tokens burned! 🔥
People and companies are heavy users of open models on the Helmcode platform.
To celebrate, we're sharing our opendata page, updated weekly.
👉 helmcode.com/opendata
And a quick thread about companies usage 🧵
English
Helmcode retweetledi
Gente no dejéis de mirar @helmcode que ofrece lo que hay en NaN para comunidad, pero enterprise-grade y en esteroides
Cristian Córdova 🐧@barckcode
@TheDevilOps Lo tengo que probar primero pero si me convence en lugar de GLM será Kimi K3 quien disponibilicemos para empresas si
Español
Want to understand what we do at Helmcode, how inference works, and where open models are heading? @Saulgmz_ covered all of it in one video-blog, including an interview with our CEO & Founder @barckode.
He also took a 9-hour bus trip to join our first in-person event, just to make it. He's the GOAT 🐐
The community we're building at nan.builders is insane. This is the proof. Saul deserves to be our first NaN Ambassador :)
Watch the video, like, and subscribe to his channel! (Vídeo it's in spanish)
youtube.com/watch?v=F4bFmY…
YouTube
English
Helmcode retweetledi
Dos semanitas en @helmcode. Hemos venido a JUGAR 😏
He escrito un pequeño post de presentación de quién soy y qué vengo a hacer.
"Mi rol aquí mismo es curioso, porque no puedo generar contenido técnico al nivel necesario. Así que soy una especie de minero y editor del conocimiento. El conocimiento técnico ya existe en el equipo: en los problemas que se solucionan, en las preguntas de la comunidad, en los datos o en las llamadas con clientes. Mi trabajo es pulir los diamantes en bruto que me encuentro."
Si quieres saber lo que voy a hacer en Helmcode y en NaN o quieres conocerme un poco más, te dejo 5 min de lectura 👇
helmcode.com/es/posts/hello…
Español
Introducing Image Generation on Helmcode for Enterprise accounts 🖼️
🎨 Text-to-image powered by flux-2-klein
🖥️ Generate from your dashboard or via API
✨ Prompt enhancement on by default (better results with zero extra work) :)
Already active in all Enterprise accounts, no setup needed.
Read the docs → #tag/images" target="_blank" rel="nofollow noopener">helmcode.com/docs/api#tag/i…
English
The most-used model on our platform was crashing once an hour after we migrated it to NVFP4.
Everything pointed to vLLM. Three days of investigation, a GPU core dump and a cuda-gdb session later: the bug was in cuDNN, already fixed by NVIDIA... in a version torch doesn't ship yet. The final fix was one line.
The result was worth the trip: +11% throughput and -12% latency per output token on the same GPU. And along the way, we contributed the full diagnosis upstream to the @vllm_project.
This is a deep dive written by @barckcode about how our technical team optimized @Alibaba_Qwen 3.6 and debugged an issue affecting our users' community.
Helmcode@helmcode
English
A picture recap of our first in-person event! 🔥
Last week we teamed up with @dobleio and @Startup_VLC for a deep dive into agentic engineering. Álex delivered an absolute masterclass.
Packed room, great questions and people left buzzing with ideas about how to improve their workflow.
If you missed it, we have all the slides for you here: doble.io/e0b59cc146a957…
Thanks to everyone who joined! More coming soon
Where we should be next? 👀
English
New docs on the block! 📖🤓
We've rebuilt the Helmcode docs from the ground up.
⚡Quickstart in minutes
🤖 Full model catalog + agents guide
🇪🇺 EN & ES
Built with @scalar 's OSS docs library by @Cris_gl19
Go! → helmcode.com/docs
English
Slides from @dobleio's talk on agentic engineering at our first event in Valencia are up!! 🔥
Worth reading if you're moving agents from demo to production. The best talk on agent engineering we've ever heard! 👇
Alejandro Vidal@dobleio
This week I gave a talk in Valencia on serious agentic engineering from first principles! doble.io/e0b59cc146a957… The through line: AI changes how we build, but it also changes the economics of what's worth building. Converting tokens into value is not trivial and can be optimised. Product engineering is now about managing the tradeoff between tokens, cost, human attention and code production. We need new tools: nested feedback loops at every frequency, user behaviour monitoring, cognitive debt management, and more. We need to catch misalignment early, because errors compound and agents copy what they see. And we need to spend humans where they're scarce: upstream, on design, on types, interfaces and state machines, not on reading infinite diffs. Exciting times for software engineering! Thanks to @helmcode, @Startup_VLC and @vlcinnovation for hosting, and to @borjaperfra, @jlhortelano, @barckcode and @ptrujillano for making it happen.
English