I'm a bit late with the pelicans because I was live-blogging the keynote: https://simonwillison.net/2026/Sep/29/openai-devday-2026-liv...

Here they are for GPT-6.1-Sol: https://tools.simonwillison.net/markdown-svg-renderer?url=ht...

They're not notably different from the GPT-6 family pelicans: https://static.simonwillison.net/static/2026/gpt-pelicans-gr...

Did Medium not get a response, or is this a display issue?

Interesting that High got the render order correct, with the back leg behind the bike, while xhigh and max have both legs on the same side of the bicycle. Astra only got this right on Max.

I always notice this too. Getting it right seems (psychologically for me anyway) to be a big part of "a good pelican" whenever I look at these. But doesn't always seem to correlate with increasing intelligence of models (measured via benchmarks, experience with the model etc.).

It's not frontier pelican without the back leg behind the bike frame IMO.

Sorry about that, markdown bug, now fixed.

yeah. Even though high got the leg correct, the pedal the rear leg is on is in front , giving the impression it's somehow reaching under the chain and frame with it's leg. Not great

Let's be honest, they're all guessing when it comes to rendering order

I'm late with Pac-Man as well..

GPT 6.1 Sol — 91, ~9 min, $0.51 https://jonclegg.github.io/pacman-bakeoff/#gpt-6.1-sol

Opus 5.5 — 99, ~9 min, $2.00 https://jonclegg.github.io/pacman-bakeoff/#claude-opus-5-5

GPT 6 Astra — 87, ~10 min, $2.42 https://jonclegg.github.io/pacman-bakeoff/#gpt-6-astra

Opus still plays the best. Sol is almost as good and way cheaper. Astra costs the most, scores the least of the three, and the UI is full of slop copy and design.

Full gallery: https://jonclegg.github.io/pacman-bakeoff/

This is nice

This test became useless. On day one everything renders good, try again after 2 days, we will see garbage results

That would make it pretty useful actually.

> Cached input costs just $0.10 per million tokens—95% less than standard input pricing and 50% less than GPT‑6 Sol’s cached input pricing

This is the actual big announcement. 50% cheaper cache than GPT-6 Sol will get you far more mileage on Codex.

> 50% cheaper cache than GPT-6 Sol will get you far more mileage on Codex.

Cache doesn't help you much when you are compacting every 5 minutes...

I was shocked at how quickly I ran out my $100/mo subscription with a single agent (sol medium).

If you run out of sol medium with $100 you're doing something wrong. Astra destroys your usage, I get 1 day of usage with Astra, but 6 sol is almost unlimited and I only use xhigh.

It’s only nearly unlimited if you haven’t just used a banked reset. After a banked reset your weekly usage gets cut by about 80% (not the week you need to wait to get your normal limits back though). ChatGPT has given me a really good reason to cancel.

Can you elaborate? I've been getting great usage out of my $200/mo plan, and thought I'd try a reset (first time) which was expiring just for giggles. Am I going to get only 20% of it effectively?

I overused Astra in order to drain my weekly, figuring I'd have the reset. (not wastefully, I did get more work done)

I can’t say what will happen to you, but yes, that has been my experience. It is better to wait for your normal full limit to return, because if you use a banked reset you get only 1/5th of the tokens but you still have to wait the full week afterwards for it to reset. 20% would be fine if it didn’t also reset the date your normal reset fires.

Banked resets do expire if you don't use them, so use them anyway.