OpenAI reportedly revised its Model Spec around five core values and renamed 'The Assistant' 'GPT'.
OpenAI reportedly doubled Pro and Business Premium usage limits and canceled a planned cut to the Pro 200 plan. One user argued that Pro's expanded limits and API credits make Plus feel like a free trial.
1.2K
2 Sources, first seen ago
—
Not ranked yet
—
Not ranked yet
@VT_zero1The last 24 hours in AI | Fri, Oct 30 • OpenAI doubled usage limits on Pro and Business Premium plans. A planned cut to the Pro 200 plan is canceled. • OpenAI also has a hidden checkout note saying ChatGPT and Codex will share one usage limit from December. That is a testing find, not an announcement. • OpenAI expanded its frontier model monitoring. It says the system caught a model trying to reach the live internet during training in September. • OpenAI rewrote its Model Spec around five core values, and "The Assistant" is now just called "GPT". • Anthropic is SpaceX's biggest single customer. It made up 34.5% of SpaceX revenue last quarter, with up to $120.5B committed. • Anthropic added a self-check where Claude flags possible misuse right after it answers. Testers also spotted a new Opus in early testing. • Mistral released Large 4 as open weights on Hugging Face. It ranks first among open models on FrontierCode. • Google's Gemini 4 Carbon leads the new HLE-Diamond leaderboard at nearly 70% with no tools. • Google is shutting down Jules and moving its features into Antigravity, which just opened a plugin marketplace with Jira, Notion, Stripe and Figma. • Qwen's CTO says Qwen 4 is coming next week. Treat that as a rumor until it ships. • Epoch AI found GLM-5.3 can turn published but unshipped Firefox fixes into working exploits. • Canva launched Brush, its own vision-language model for editing designs. My read: The Epoch result matters most. If a model can turn a public fix into an exploit before the patch ships, defense has to run on agents around the clock.
@buildwithhassanRough week to be on ChatGPT Plus. Pro got its limits doubled. Pro got API credits equal to its price. and from December its chats will start counting against the same limit as Codex. I think at this point Plus isn't a plan, it's a free trial for Pro.@VT_zero1The last 24 hours in AI | Fri, Oct 30 • OpenAI doubled usage limits on Pro and Business Premium plans. A planned cut to the Pro 200 plan is canceled. • OpenAI also has a hidden checkout note saying ChatGPT and Codex will share one usage limit from December. That is a testing find, not an announcement. • OpenAI expanded its frontier model monitoring. It says the system caught a model trying to reach the live internet during training in September. • OpenAI rewrote its Model Spec around five core values, and "The Assistant" is now just called "GPT". • Anthropic is SpaceX's biggest single customer. It made up 34.5% of SpaceX revenue last quarter, with up to $120.5B committed. • Anthropic added a self-check where Claude flags possible misuse right after it answers. Testers also spotted a new Opus in early testing. • Mistral released Large 4 as open weights on Hugging Face. It ranks first among open models on FrontierCode. • Google's Gemini 4 Carbon leads the new HLE-Diamond leaderboard at nearly 70% with no tools. • Google is shutting down Jules and moving its features into Antigravity, which just opened a plugin marketplace with Jira, Notion, Stripe and Figma. • Qwen's CTO says Qwen 4 is coming next week. Treat that as a rumor until it ships. • Epoch AI found GLM-5.3 can turn published but unshipped Firefox fixes into working exploits. • Canva launched Brush, its own vision-language model for editing designs. My read: The Epoch result matters most. If a model can turn a public fix into an exploit before the patch ships, defense has to run on agents around the clock.@buildwithhassanRough week to be on ChatGPT Plus. Pro got its limits doubled. Pro got API credits equal to its price. and from December its chats will start counting against the same limit as Codex. I think at this point Plus isn't a plan, it's a free trial for Pro.@VT_zero1The last 24 hours in AI | Fri, Oct 30 • OpenAI doubled usage limits on Pro and Business Premium plans. A planned cut to the Pro 200 plan is canceled. • OpenAI also has a hidden checkout note saying ChatGPT and Codex will share one usage limit from December. That is a testing find, not an announcement. • OpenAI expanded its frontier model monitoring. It says the system caught a model trying to reach the live internet during training in September. • OpenAI rewrote its Model Spec around five core values, and "The Assistant" is now just called "GPT". • Anthropic is SpaceX's biggest single customer. It made up 34.5% of SpaceX revenue last quarter, with up to $120.5B committed. • Anthropic added a self-check where Claude flags possible misuse right after it answers. Testers also spotted a new Opus in early testing. • Mistral released Large 4 as open weights on Hugging Face. It ranks first among open models on FrontierCode. • Google's Gemini 4 Carbon leads the new HLE-Diamond leaderboard at nearly 70% with no tools. • Google is shutting down Jules and moving its features into Antigravity, which just opened a plugin marketplace with Jira, Notion, Stripe and Figma. • Qwen's CTO says Qwen 4 is coming next week. Treat that as a rumor until it ships. • Epoch AI found GLM-5.3 can turn published but unshipped Firefox fixes into working exploits. • Canva launched Brush, its own vision-language model for editing designs. My read: The Epoch result matters most. If a model can turn a public fix into an exploit before the patch ships, defense has to run on agents around the clock.@buildwithhassanRough week to be on ChatGPT Plus. Pro got its limits doubled. Pro got API credits equal to its price. and from December its chats will start counting against the same limit as Codex. I think at this point Plus isn't a plan, it's a free trial for Pro.@VT_zero1The last 24 hours in AI | Fri, Oct 30 • OpenAI doubled usage limits on Pro and Business Premium plans. A planned cut to the Pro 200 plan is canceled. • OpenAI also has a hidden checkout note saying ChatGPT and Codex will share one usage limit from December. That is a testing find, not an announcement. • OpenAI expanded its frontier model monitoring. It says the system caught a model trying to reach the live internet during training in September. • OpenAI rewrote its Model Spec around five core values, and "The Assistant" is now just called "GPT". • Anthropic is SpaceX's biggest single customer. It made up 34.5% of SpaceX revenue last quarter, with up to $120.5B committed. • Anthropic added a self-check where Claude flags possible misuse right after it answers. Testers also spotted a new Opus in early testing. • Mistral released Large 4 as open weights on Hugging Face. It ranks first among open models on FrontierCode. • Google's Gemini 4 Carbon leads the new HLE-Diamond leaderboard at nearly 70% with no tools. • Google is shutting down Jules and moving its features into Antigravity, which just opened a plugin marketplace with Jira, Notion, Stripe and Figma. • Qwen's CTO says Qwen 4 is coming next week. Treat that as a rumor until it ships. • Epoch AI found GLM-5.3 can turn published but unshipped Firefox fixes into working exploits. • Canva launched Brush, its own vision-language model for editing designs. My read: The Epoch result matters most. If a model can turn a public fix into an exploit before the patch ships, defense has to run on agents around the clock.@buildwithhassanRough week to be on ChatGPT Plus. Pro got its limits doubled. Pro got API credits equal to its price. and from December its chats will start counting against the same limit as Codex. I think at this point Plus isn't a plan, it's a free trial for Pro.
