I’m seeing really good reviews of OpenAI’s new image model, but I find that even with Thinking, it hallucinates quite a few data points, which, therefore, makes the result unusable no matter how readable and well-formatted it is… Am I missing something?
@dotcsv
-
Claude Design: Game-Changing Animation Tool from Anthropic
By
–
NEW VIDEO in the LAB! Today we're talking about Claude Design, which for me is the most interesting announcement from Anthropic this past week. For me, it's a total game-changer that completely transforms my way of working and creating animations! I break it all down in
-
Files Deleted When Token Limit Exceeded: A Critical Issue
By
–
The combo: files deleted at the same time as it's burned through all my tokens
-
Stop Pushing Vibro-Coded Features to Production
By
–
The exact same thing has happened to me. Stop pushing to production features vibro-coded!
-
AI Tools Accessibility Erodes Competitive Advantages Rapidly
By
–
Moreover, even learning to use these tools doesn't give you any moat. Because their use becomes simpler and simpler every time. The person who starts using these tools today will encounter a much simpler learning curve and will be able to do much more with the tools than the one
-
AI Democratization Flattening Digital Economy Skills Floor
By
–
Today I'm finding myself a victim of AI psychosis, watching as we're all witnessing a flattening of the digital economy. We're all starting to be able to do everything: websites, apps, animations, designs, images, videos, music… The floor is rising, and one might think that
-
Anthropic’s Feature Presentations: Repackaging Existing Model Capabilities
By
–
100% de acuerdo con esto. Muchas de las funcionalidades que Anthropic ha ido presentando, y que dan esa impresión de avances cada día, realmente son re-empaquetamientos de capacidades que realmente ya estaban ahí, implícitas en los modelos, adaptadas para público general. Es
-
Claude Design: AI-powered design tool with style guides
By
–
Nueva funcionalidad Claude Design para hacer diseños con Claude usando como inputs tus manuales de estilo, referencias, etc!
— Carlos Santana (@DotCSV) 17 avril 2026
Justo muy parecido a una funcionalidad que me implementé la semana pasada para hacerme animaciones. Si tengo tiempo os la enseño la próxima semana. https://t.co/sQddlSUq8wNueva funcionalidad Claude Design para hacer diseños con Claude usando como inputs tus manuales de estilo, referencias, etc! Justo muy parecido a una funcionalidad que me implementé la semana pasada para hacerme animaciones. Si tengo tiempo os la enseño la próxima semana.
-
OpenAI Launches GPT-ROSALIND for Drug Discovery Research
By
–
Ojito con un nuevo modelo frontera de OpenAI para investigación en biología, medicina y descubrimiento de fármacos: GPT-ROSALIND ❤️🧬 https://t.co/6QlEmcUCYI
— Carlos Santana (@DotCSV) 16 avril 2026Heads up with a new frontier model from OpenAI for research in biology, medicine, and drug discovery: GPT-ROSALIND
-
Opus 4.7 token efficiency claims questioned
By
–
Luego es gracioso que te ponen una gráfica donde se muestra claramente cómo Opus 4.7 logra mejores resultados para un mismo uso de tokens, para luego tener que indicar en el pie de gráfica que esto no es indicativo del uso de tokens del modelo