Brian Peña Calero (@brianmsm) profile picture

Studying Máster Metodología en Ciencias del Comportamiento y de la Salud - UCM. Interest in mental health issues, psychometric and empirical research.

Madrid, Spain
Joined January 2014
Photo by Brian Peña Calero (@brianmsm)Photo by Brian Peña Calero (@brianmsm)Photo by Brian Peña Calero (@brianmsm)Photo by Brian Peña Calero (@brianmsm)Photo by Brian Peña Calero (@brianmsm)Photo by Brian Peña Calero (@brianmsm)Photo by Brian Peña Calero (@brianmsm)Photo by Brian Peña Calero (@brianmsm)Photo by Brian Peña Calero (@brianmsm)Photo by Brian Peña Calero (@brianmsm)
Brian Peña Calero retweeted
github.com/mmnavarr/overseer is the must have plugin for Tern
5
12
1
185
15,192
Brian Peña Calero retweeted
My old Echo Dot run Hermes now, instead of Alexa. I built a Home Assistant integration that connects it to a Hermes profile, with its memory and tools. Code, setup skill and the hardware recovery notes are public. github.com/majorminorlabs/he… @NousResearch
Please definitely dm me or send me your feedback - I reaaaaaaaaaaaaaaaaaaally wanna replace Google Home/Alexa for people one day.
94
101
31
1,226
177,420
Brian Peña Calero retweeted
opus'a ucuz bir işçi eklemek işi ucuzlatmıyor. reddit'te bir kişi 12 zor görevi ölçtü. opus tek başına %81 başarı oranı ile görev başına 1,20 dolar harcadı. opus planlayıp deepseek v4.1 flash kod yazınca %67 başarı oranı ile 1,29 dolar harcadı. 7 takım tasarımı, 41 kurulum, yaklaşık 1.200 koşu ile test edilmiş. opus'un başında olduğu en iyi takım %69'da kalmış. sebep model seçimi değil görevler arası devir teslim. worker kod yazarken opus'un prompt cache'i düşüyor ve opus aynı contexti yeniden okuyor ve tekrar maliyet çıkartıyor. bu kalem tek başına worker'ın maliyetinden fazla çıkmış. aynı tabloda gpt-6.1 terra %78 ve sol %75 başarı oranı ve görev başına 0,17 dolar olarak ölçülmüş. genel olarak 12 görev kesin bir karar vermek zor ve küçük farklar gürültü yaratabilir. ama 41 takımın hiçbiri tek opus'u geçmemiş.
24
17
6
194
28,571
Brian Peña Calero retweeted
We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math. It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training. We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵 Blog: rikiyat.github.io/blog/weigh…
Readers added context they thought people might want to know
LLM was given weights explicitly as text in prompt and used code/forward passes (per authors' logs) to ablate Qwen looping direction, a known thinking-mode issue per model card. Primarily fixed repetition in that mode, not general 'natively' improving weights/math. rikiyat.github.io/blog/weight-sp… github.com/RikiyaT/lm_und… huggingface.co/Qwen/Qwen3.5-0… huggingface.co/RikiyaT/Qwen3.…
2,362
7,275
5,731
55,922
9,865,140
Brian Peña Calero retweeted
Did you know the same intelligence can cost 10× more, need 6× more output tokens and take twice as long? In addition to task spend, Agent Arena now shows which agents give you best performance for every output token generated and every minute you wait. These real-world usage plots reveal that costs grow most steeply with intelligence, waiting longer pays off unevenly, and output tokens barely predict the bill. To find the best model for you, see the link in thread.
30
7
5
227
29,205
Brian Peña Calero retweeted
Introducing Phonic-1, the most advanced speech-to-speech model for agentic tasks — from @phonic_co, the speech-to-speech model company. We’ve had a breakout year and raised a $27.5M Series A led by @firstround with participation from @Lux_Capital, @felicis, and Definition.
147
193
135
1,263
1,087,869
Brian Peña Calero retweeted
From recognizing text to understanding documents. 📄 Meet Xiaomi-OCR-0, an open-source document understanding model with just 0.8B parameters. It brings document parsing, key information extraction, and OCR-oriented visual question answering into one lightweight model. It scores 95.24 on Real5-OmniDocBench and 83.2 on average across five OCR-oriented VQA benchmarks. On OmniDocBench v1.6, Xiaomi-OCR-0 ranks first in document parsing performance among models with fewer than 1B parameters. Model weights, inference code, an interactive demo, MCP service, and Agent Skill are now open source. 👇 🤗 Huggingface: huggingface.co/xiaomi-resear… 🔗 GitHub: github.com/xiaomi-research/X… 📄 Demo: huggingface.co/spaces/loongw… #XiaomiAI #OCR #OpenSource
58
311
54
2,863
153,085
Brian Peña Calero retweeted
Montar llamadas VoIP para un agente de IA en España es un rollo. Twilio te pide números, verificaciones y papeleo. Las APIs de WhatsApp y Telegram tampoco te lo ponen fácil. Y al final pagas por minuto por algo que solo quieres para ti. Así que me he hecho una app que lo resuelve: agentring. Le da a tu agente la capacidad de llamarte al móvil, y a ti la de llamarle a él. Cómo funciona: → Tu agente (Claude Code, Codex, OpenClaw, n8n o un script) tiene una herramienta "call". Cuando necesita que decidas algo, la usa: "la migración ha fallado, ¿hago rollback o lo arreglo y reintento?". → Tu móvil suena como una llamada normal, aunque esté bloqueado. Coges y una IA de voz te explica qué pasa. → Lo habláis, tú decides, y tu agente recibe la conversación escrita y sigue trabajando con tu respuesta. → Al revés también: abres la app, pulsas "Llamar al agente" y la IA le pregunta a tus agentes lo que necesites mientras habláis. → Si no lo coges, te llega como mensaje y respondes desde la notificación. Yo ahora mismo lo uso para que me llame si: • Sale en Twitter o en alguna fuente oficial una vulnerabilidad crítica que me afecte. • Detecta algún problema de seguridad en mis servidores. • Entra un correo crítico. • Se cae algo. Con todo lo que está pasando, creo que hace falta tener agentes levantados revisando tu seguridad, igual que va a haber cientos de miles intentando entrar. Y si encuentran algo, un mensaje no basta: tiene que sonarte el móvil. Por dentro: • La llamada va por VoIP desde tu servidor a tu móvil. Sin números de teléfono, sin Twilio y sin pagar por minuto. • La voz es Gemini Live: muy barato, y en AI Studio tienes capa gratuita para probarlo. • Es open source y lo montas tú, en un VPS o en tu propio ordenador. Se conecta a tu agente en un par de minutos. Por ahora es solo Android. ¿Qué se os ocurre montar con esto? Os dejo el repo en el siguiente comentario por si alguién lo necesita 👇
40
42
8
450
34,118
Brian Peña Calero retweeted
No digo nada pero Qué infravalorado está lo que ha hecho @mrm8488 con huggingface.co/mrm8488/embed… gracias por la ideaza, el provecho que le voy a sacar no tiene ni sentido ^^
5
10
134
8,843
Brian Peña Calero retweeted
Day 25/ We are doubling all usage limits for all Pro 100/200/500 tiers and Business Premium, effective today. We heard your kind feedback and won’t be reducing usage limits for Pro 200, so your Work and Codex usage will stay the same. We’re also going to continue offering extended limits for dots for now, and usage of your dot will still continue to be separate from your default usage. And we'll of course also keep shipping better models and other efficiency improvements that increase your effective usage over time.
7,404
2,703
3,649
39,186
7,498,490
Brian Peña Calero retweeted
claude designed a 3d part for my robotics experiment in freecad. i'd never used cad, so every time i wanted to change something i couldn't even say which part i meant so i gave my coding agent its own cursor. i circle the thing, ask, and it points back
3
4
13
1,287
Brian Peña Calero retweeted
I did say we'd be releasing our models periodically...introducing Gemini 4 Carbon, our next frontier model. Following Argon's release just two weeks ago, Carbon shows the pace at which we’re pushing the frontier, it ranks #1 on the recently updated v5.0 of @ArtificialAnlys Intelligence Index at 59. It also leads across key cybersecurity evaluations scoring 88.2% on SEC-bench Pro for finding vulnerabilities, 54.8% on ExploitGym for exploitation, and 71.0% on CWE-bench for patching. We’re seeing significant improvements in long-horizon tasks even compared to Argon, and we’ve strengthened frontier safeguards. As we did with Argon, we’re going to roll it out gradually + responsibly. Some trusted cyber defenders will get access starting this week. We'll be expanding it to more partners through our Fairwind Program, ahead of wider availability soon, so hang tight!
2,082
2,742
1,919
34,062
6,959,494
Brian Peña Calero retweeted
I don't code anymore, but I miss looking at code. Yesterday I threw together a TUI that reminds me of Neovim, that watches the agents work. github.com/bjarneo/shoulder
29
9
3
289
20,766
Brian Peña Calero retweeted
OpenDisplay for Linux is now available, Sender & Receiver! It's still early, therefore happy to receive feedback and contributions to support as many linux setups as possible.
24
100
11
1,570
76,876
Brian Peña Calero retweeted
Introducing Real World Engineering Bench (RWE-Bench). AI isn’t great at designing rockets, engines, or chips yet. With better evals and RL environments, it will be. Software was just the beginning. The physical world is next.
87
152
58
1,232
161,051
Brian Peña Calero retweeted
🇵🇪 Ya está en línea: el Censo 2017 vs el Censo 2025, distrito por distrito. 👉 censo-2025.appsbi.net Empiezo con un dato que parece contradictorio. 🚗 En 2025 hay 1,285,772 hogares con auto o camioneta, 196,157 más que en 2017 (+18 %). Pero el porcentaje bajó de 13,2 % a 12,2 %. ¿Por qué? Los hogares crecieron más rápido que los autos: pasamos de 8,25 a 10,57 millones de hogares en solo 8 años. Hay más familias, más pequeñas. 📍 Y la brecha es enorme: • San Isidro: 70,9 % de hogares con auto • San Borja 64,1 %, La Molina 62,9 %, Miraflores 62,2 % • En distritos de Loreto como Jenaro Herrera, Yaguas o Yaquerana: 0 % Dos realidades en nuestro mismo pais. 📊 Qué puedes hacer en la app: ✅ Comparar 60 preguntas que se hicieron en ambos censos: auto, refri, internet, celular, SIS, EsSalud, agua, desagüe, piso de tierra, gas natural, leña, idioma materno, educación y más ✅ Verlas en el mapa de los 1,893 distritos: 2017, 2025 o cuánto cambió ✅ Ver cantidades y %, y el top 20 de distritos que más subieron y más bajaron ✅ Bajar a tu departamento, provincia o distrito ✅ Explorar todas las variables del Censo 2025 🔍 Por qué no es un simple cruce: El INEI cambió códigos y opciones de respuesta entre un censo y otro. Emparejamos pregunta por pregunta y documentamos qué se agrupó y qué no es exactamente comparable. Además, validamos que cada distrito cuadre con el total de su región, y los totales 2017 coinciden con los oficiales del INEI. 🔓 Todo es abierto: • Datos en CSV para descargar • Metodología completa • Código completo para clonar: extractores, endpoints del INEI, backend y frontend ℹ️ Los datos son de la población y los hogares censados (32,7 millones de personas), los mismos que el INEI publica en su página. Los 34,2 millones de habitantes que anunció el INEI suman además la población omitida que estima su encuesta postcensal. Cualquier mejora es bienvenida. ¿Encontraste un error, una pregunta mal emparejada o una idea para la app? Comenta y la implementamos. 📥 Y si quieres hacer tus propios análisis, eres libre de hacerlo: toda la información es pública. Descarga los CSV, la base y el código en 👉 censo-2025.appsbi.net/descar…
29
279
6
905
40,289
Brian Peña Calero retweeted
Flint makes a great teacher. Ran it locally on a 24GB M5 Pro, had it write + double-check 1.2k text-to-SQL pairs, then distilled a 331MB model with 0.2s per query (Flint: 7.4s). Model, data + demo: huggingface.co/datasets/alex…
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.
3
5
89
4,811
Brian Peña Calero retweeted
¡He creado una biblioteca para usar IA en tu navegador! Modelos de pesos abiertos y especializados. ✓ Gratis y Open Source ✓ Sin APIs ni servidores ni pago por tokens ✓ Tus datos nunca salen de tu dispositivo Se llama runonweb: → github.com/midudev/runonweb.…
66
343
13
3,047
132,624
Brian Peña Calero retweeted
Introducing epstein.md 👋 A 927,400+ Character .MD File that's specifically trained to skip every vibe coded AI Slop design principle. - no box shadows - no emojis - no '🟢 LIVE ' badges - 0 blue/purple gradients & religiously ignores 350 most common AI words (delve, vibe, seamless ...) It's the result of 11 months of 8 hrs/day Twitter doomscrolling and constant additions to the codebase of our vibecoding ui. Wholeheartedly hoping this finds the most design-needy indie hackers.
136
86
81
2,115
446,950
Brian Peña Calero retweeted
Introducing GLM-Legal Built on GLM-5.3-Flash, GLM-Legal matches frontier models on mainstream legal benchmarks. Designed for real legal workflows: contract review, evidence analysis, compliance, and more. Tech Blog: z.ai/blog/glm-legal Apply for API Access: z.ai/consultation/legal Try AutoClaw: autoclaw.z.ai
356
574
323
7,784
866,888