Ultrafast is rolling out today for GPT-6.1 Sol in the API, Codex, and ChatGPT Work. Near-Astra intelligence at up to 8x faster speeds than Sol Standard, so you can build as fast as the ideas come.

Oct 8, 2026 · 6:26 PM UTC

468
369
379
5,404
1,991,384
RelevantRecentLikes
API pricing for GPT-6.1 Sol in Ultrafast mode is $12 per million input tokens and $60 per million output tokens. Built for work where speed and intelligence make a difference: debugging an outage, agents navigating apps, and live experiences where every second counts.
Image from OpenAI Developers (@OpenAIDevs)
33
31
41
619
128,770
In Codex and ChatGPT Work, access is available on Pro 500, eligible usage-based Enterprise, and credit-based Edu plans. Enterprise admins must enable access. Ultrafast for GPT-6.1 Sol is available in all supported regions, including support for data residency in the US and EU. We’ve also added support for EU data residency for GPT-6.1 Sol Fast and GPT-6 Luna Fast. developers.openai.com/api/do…
Ultrafast mode | OpenAI API

Ultrafast mode | OpenAI API

Use Ultrafast mode with GPT-6 Astra and GPT-6.1 Sol over WebSockets, with SDK examples and an HTTP alternative.

developers.openai.com
46
18
26
344
95,073
only for Pro 500 btw
2
60
7,285
Is this already counting @thsottiaux s 1.5x speed up announcement or is this on top of that?
2
32
4,860
nice now i can use sol ultrafast!
Image from 💺 (@patience_cave)
2
31
6,178
The $200 plan has become worth a fucking shit
1
2
234
13,530
Looking forward to ultrafast being the default
3
21
4,059
so you mean regular speed but at the cost of ultrafast
14
2,265
@OpenAIDevs
Image from Coinbase Predict (@CoinbasePredict)
1
1
10
2,716
insanity
11
2,042
dang rip 14x we never even got to know ya
3
1,366
Really excited to test this out.
1
1,615
ok, let me test it!
611
xitter.twitterapi.xyz/nicoritschel/status/21…
Nico Ritschel (@nicoritschel) profile pictureNico Ritschel
@nicoritschelOct 8
Replying to @CodexReleases
No one cares unless OpenAI is paying for the tokens
228
stop talking nonsense you downgraded all models, GPT is stupid af I'm not a dev, but I say how to take certain actions and then SOL or Astra agrees with me and appologizes, continuously. I thought it's isolated case, but it's rule, with more people reporting same issues 👎
317
First you dilute the $200 plan by half, then launch features only for the $500 plan? This is Black Mirror-level shit.
1
1
96
7,404
They made the taxi faster and somehow the meter won.
2
1
26
5,053
why only Pro 500?... I get astra cause its has more usage, but 6.1 Sol?
24
4,284
Considering the fact that 6.1 was performing at around ~30 TPS with fast mode enabled over the last 2 days, we can expect a maximum of around ~120 TPS on ultra-fast mode, which ties with opus 5.5's speed in standard mode. Nice update, thanks!
3
21
3,401
incredible… even $200/month isn’t enough anymore
19
2,564
since you downgraded pro 200, you could have at least let it available for 200 pro users
8
1,253
I don't think anyone wants this. I never wanted it. I just wanted the regular models to work reliably. They didn't. They churned all my tokens on extraneous work and rarely achieved the objectives. Idk what happened after gpt-5.5, but it was overall a turn for the worse.
4
9
3,127
When will it be available for Pro 200 users?
7
1,879
Eight times faster and still no model that admits it is wrong
6
1,056
still expensive, but finally makes sense to use it
6
2,821
Already showing in ChatGPT Work for me. Standard, Fast, Ultrafast, right in the effort picker. Before you leave it on: on Pro it eats your included usage at 8x the Standard rate. Bought credits go 6x. learn.chatgpt.com/codex/agen…
6
3,248
Nothing for $200 Pro? Disappointing
4
1,077
Churned to Anthropic once more thanks
4
623
is this video made by Opus 5.5?
3
1,100
I would buy $500 if it wasn't only 25x. I have a pro sub that is yet in 20x so it would be a huge loss buying that sub.
1
3
4,123
Sol after ultrafast
3
2,112
ultrafast in codex is pro 500 only, and pro 200 drops from 20x to 10x plus usage after oct 29. gamers know this patch. nerf the base game, sell the speed boost
3
1,236
Disappointing
Donno (@donnodontaskme) profile pictureDonno
@donnodontaskmeOct 9
Replying to @IndependentEco
Really upset about this for not being open and honest they can't offer on lower models, unless u dig deeper u can't find which model it's available. They can keep price same and stuff worst case my tokens I burnt the way I want to be . Not giving access is bad imop.
2
1,021
for 500$ subscribers only... you forgot to mention XD You’re turning into a company that fleeces the wealthy at the expense of those who can’t afford it... you’re really having a blast! Congratulations! The "Open" in your name now refers only to your wallets.
2
343
near-Astra smarts at $60 per million out, $10 more than Astra itself
2
1,442
xitter.twitterapi.xyz/sughanthans1/status/21…
Sughu (@sughanthans1) profile pictureSughu
@sughanthans1Oct 8
dots aren't special because of the model. they're special because they get a computer that stays on. log into a browser once and it remembers. my codex needs my laptop open. close the lid and my agent goes to sleep. grokbot has its own computer too and i can reach it from anywhere. that's the whole game: always on, reachable from anywhere, data that stays.
2
463
Load more replies