17 July 2025
Topic: Models And Releases
ChatGPT agent System Card
ChatGPT agent System Card: OpenAI’s agentic model unites research, browser automation, and code tools with safeguards under the Preparedness Framework...
TL;DR
OpenAI's agentic model, ChatGPT, has integrated research, browser automation, and code tools under the Preparedness Framework
Key Takeaways:
Key Takeaways:
- Provides a unified platform for research, automation, and coding with built-in safeguards.
- Leverages the Preparedness Framework to ensure the responsible advancement of language and agentic models.
- Combines the strengths of agentic, browser automation, and coding tools within a single framework.
Agent bio bug bounty call
OpenAI invites researchers to its Bio Bug Bounty. Test the ChatGPT agent’s safety with a universal jailbreak prompt and win up to $25,000....
TL;DR
OpenAI invites researchers to its Bio Bug Bounty, offering up to $25,000 for testing ChatGPT agent's safety with a universal jailbreak prompt.
Key Takeaways:
Key Takeaways:
- Researchers can test ChatGPT's safety with a universal jailbreak prompt.
- OpenAI is offering up to $25,000 for participating researchers.
- The Bio Bug Bounty aims to improve the safety of ChatGPT, the company's popular chatbot agent.
Seems like OpenAI is planning to release Agent Mode, codenamed “Odyssey”. (check all 5 pics)
support for Ernie 4.5 MoE models has been merged into llama.cpp
Agent = Deep Research + Operator. Plus users: 40 queries/month. Pro users: 400 queries/month.
AI models are getting dumber?
My simple test: Qwen3-32b > Qwen3-14B ≈ DS Qwen3-8 ≳ Qwen3-4B > Mistral 3.2 24B > Gemma3-27b-it,
No More Automatic Re-Try