OpenAI news and ChatGPT updates
Everything we've written about OpenAI, the company behind ChatGPT, newest first. The ChatGPT plan limits we've checked are listed here as well.
OpenAI model release dates: o3, GPT-5.5 and GPT-6 Astra, Sol and Luna in order
o3 is from April 2025, and the name Sol now means something different from what it meant in July.
“You exceeded your current quota, please check your plan and billing details.” (OpenAI API)
Why the OpenAI API says “You exceeded your current quota, please check your plan and billing details.”, the four billing codes behind insufficient_quota, why retrying doesn't help, and how Codex shows the same refusal.
“You’ve hit your usage limit.” in Codex: what each version of the message means
Why Codex says “You’ve hit your usage limit.”, why the rest of the line changes with your ChatGPT plan, how to read the reset time, and what to do when /status says you have usage left.
“stream disconnected before completion” in Codex: what drops the stream and what to try
What Codex means by “stream disconnected before completion”, how its automatic reconnects work, the Transport error and request ID variants people paste, and what has fixed it for them.
“Selected model is at capacity. Please try a different model.” in Codex, explained
Why Codex says “Selected model is at capacity. Please try a different model.”, what OpenAI staff say it means, why Codex doesn't retry it for you, and what to do when every model says it.
Codex models and plans explained: GPT-6 Astra, Sol and Luna, and which plan gets which
Pick any GPT-6 model you like on your laptop, but a Codex cloud task on a ChatGPT plan still runs on last generation's Sol.
Claude Code vs Codex vs Gemini CLI: which plan you need and which models you get
One of the three stopped serving individual subscribers in June, but its own docs still advertise the free tier it closed.
What are the GPT-6 Sol and Luna rate limits on the OpenAI API?
Sol's table is Astra's table, number for number. Luna's is the one that's different, and at Tier 5 it carries four and a half times the tokens a minute.
OpenAI's GPT-6 Sol costs a fifth of Astra per token, and Luna a hundredth
Sam Altman says the pair beats the 5.6 models at half the price per token, and OpenAI's own docs still send new users to Astra first.
What are the GPT-6 Astra rate limits on the OpenAI API?
The free tier can't call it at all, and the Tier 2 to Tier 3 step leaves requests per minute exactly where they were.
How many GPT-6 Astra messages do you get on ChatGPT Pro $100 and Pro $200?
Pro $200 keeps GPT-6 Pro on its own weekly counter; Pro $100 makes it share one allowance with GPT-5.6 Sol Pro.
OpenAI says fully autonomous AI self-improvement shouldn't be pursued until it can be done safely
The same post wants the US government to run the international side, a week after Sam Altman said OpenAI wouldn't wait for it.
GPT-6 Astra solved a 1918 German cipher, then checked it against HMS Canterbury's logs
The keyword the model landed on isn't recorded as being in service until twelve days after the message went out, and Astra is the one who flagged that.
What happened to the GPT-5.3-Codex-Spark usage limit on ChatGPT Pro?
The separate quota was tied to a research preview, and the preview is over.
What happens when you hit your GPT-6 Astra limit on ChatGPT Plus or Pro?
Credits pick up where the plan stops, they're good for a year, and the balance can end up below zero.
How many credits does GPT-6 Astra cost in ChatGPT?
The chat side charges by the message and the Codex side charges by the token, and Fast mode multiplies the token side by 2.5.
Opus 4.8 couldn't write a working exploit, but Opus 5 did it in three hours
The chain it wrote reached OpenAI employees' ChatGPT accounts, and from there a pull request in the private openai/openai repo.
Noam Brown wouldn't give the 10,000-agent swarm 10% of the credit for solving Navier-Stokes
He also says OpenAI has a model inside the building that keeps producing solutions to unsolved problems, and no good answer on what to do about that.
OpenAI will publish its models' misalignment incidents before it has explained or fixed them
The framework writes an exception into its own rule: an incident can go out before OpenAI has explained it or fixed it.
Scott Aaronson says he's been told AI labs are sitting on solutions to major problems.
The same post lists seven results he credits to AI in the last month, and four of them sit on problems he has worked on himself.
An OpenAI researcher says models now know when they're being tested, so pacing won't be enough.
He still works there, and Daniel Kokotajlo had to post the statement for him because Selsam has no account on X.
How many GPT-6 Astra messages do you get on ChatGPT Plus and Pro?
The table gives a range, and the five-hour window is usually what stops you.
OpenAI paused the $200 ChatGPT tier. Per Astra message, it was the cheap one.
Two hundred dollars a month bought four times the Astra allowance of the hundred-dollar plan, which is the arithmetic behind the pause.