OpenAI Livestream

OpenAI’s latest Livestream hub quietly nudged everyone toward its shiny ChatGPT Images 2.0 update, and the internet immediately did what it does best: pull receipts, compare notes, and argue about vibes. One sleuth shared the model’s official spec page, a.k.a. a “model card,” for the new image tool gpt-image-2, noting “API Pricing is mostly unchanged… the output price is slightly low…” Cue the price detectives. Meanwhile, another commenter said the real win is usability: if you run out of your main plan’s usage, you can still “use worse models” and tap separate image limits—plus a quirky “Nano Banana” option—making the whole subscription feel more flexible for non‑coders.

Safety folks chimed in with receipts too, dropping a system card with safety details (there’s a direct PDF), while the conspiracy crowd rolled in after someone claimed their Hacker News link to the announcement was “being penalized by the HN algorithm.” Big Tech shadow‑ban theories, assemble!

And because no launch day is complete without a meme, one user pitched “hidden message on a grain of rice” as the ultimate obfuscation flex for image prompts. So the room splits three ways: price‑watchers debating value, practical fans praising the “use it even when you’re out” safety net, and jokesters optimizing for Easter eggs. It’s peak internet energy—equal parts receipts, rants, and rice.