We first fine-tune the model to follow instructions, stay within guardrails, and keep language consistent. Then we run on‑policy RL to improve search accuracy and tool efficiency while preserving those behaviors.
GENERATIVE AI
-
New Research: SFT+RL Pipeline Boosts Search-Augmented AI Accuracy
By
–
We've published new research on how we post-train models for accurate search-augmented answers. Our SFT + RL pipeline improves search, citation quality, instruction following, and efficiency. With Qwen models, we match or beat GPT models on factuality at a lower cost.
-
Discovering Novel LLM Experts via Task-Capability Coevolution
By
–
Most LLM development still optimizes one model at a time. So this paper from Sakana AI proposes AC/DC (Assessment Coevolving with Diverse Capabilities), where models and tasks evolve together. The main idea is
-
Developer Questions Prompt Rewriting in Image Generation API
By
–
I've been mostly prompting the new model directly via the client.images.generate() API, I have no idea if that rewrites my prompts at all or if it passes them straight to the model – details here
-
ChatGPT Generates Where’s Waldo Style Raccoon Ham Radio Image
By
–
That's from this ChatGPT conversation – https://
chatgpt.com/share/69e90ba5
-7760-83e8-9e09-f06e25a90930
… – my prompt was "Do a where's Waldo style image but it's where is the raccoon holding a ham radio" -
Reverse Engineering AI Image Prompt via Browser Network Inspector
By
–
… turns out if you dig around in the browser network inspector enough you CAN find the prompt – here's the prompt it used for this image
-
Process Reward Models Grade Robot Performance Like Sports Replay
By
–
What if we could audit a robot's performance like a sports replay, grading every move, not just the final score? Researchers from Peking University, Chinese Academy of Sciences, and the Beijing Academy of AI present PRM-as-a-Judge. They use a "Process Reward Model" to watch a
-
OpenAI Workspace Agents: Automating Tasks in the Cloud
By
–
🔴 ¡OPENAI WORKSPACE AGENTS!
— Carlos Santana (@DotCSV) 22 avril 2026
Aquí está la evolución de los GPTs traídos a la era agéntica. Tareas que quieras automatizar para que un agente se ocupe trabajando en la nube.
Conéctalos a tu contexto, asígnales un rol y una tarea y listo!👇pic.twitter.com/5iuhHwvC3tOPENAI WORKSPACE AGENTS! Here's the evolution of GPTs brought to the agentic era. Tasks you want to automate so an agent takes care of them working in the cloud. Connect them to your context, assign them a role and a task, and you're set!
-
AI-Generated Eight-Panel Manga with Consistent Characters
By
–
8. Eight-panel manga sequence
Consistent characters and art style across all panels.