Automod

[iyda @notiyda]

with the release of GPT 6 Astra and now that mostly everyone has access to it do you think @thsottiaux is going to do any more resets? for those who well.... used up a lot of their limits testing this amazing new model

Hey tibo I think the people might wanted another banked reset or global reset no? ;) i think for this clearly amazing release we should celebrate on Monday with a reset don't you think?

GPT 6 Astra is genuinely an amazing and incredible new model and everyone should try it out to it's fullest capability, even on low or medium the model is still a very good model

[iyda @notiyda]

GPT 6 Astra is a really token efficient model if you thought 5.6 Sol was efficient astra is even MORE even on the max reasoning levels

It is 3x more efficient than Claude Fable 5.1

GPT 6 Astra on High is more efficient than Opus 5 on low reasoning and provide MUCH greater results

[iyda @notiyda]

GPT 6 Astra has been updated on the Artificial Analysis Intelligence Index and is now getting 2nd place just behind Claude Fable 5.1

GPT 6 Astra is also acheving the second highest score on DeepSWE just under Muse Spark 1.3

GPT 6 Astra is a great model you should try it out and let ur idea's so wild ;) most likely a week of resets and improvements to come

[iyda @notiyda]

OpenAI just removed this post

https://t.co/XjvFBpoLcB

Did they accidently leak GPT 6 Astra and it isn't suppose to come today..?

is GPT 6 Astra even going to release today... or are we all just getting played.. we shall see in 30 minutes i guess

[iyda @notiyda]

GPT 6 Astra pricing this is how it stacks up compared to benchmarks check below on my other post <3

GPT 6 Astra seems like a god sent model that i hope we are getting in the next 30 minutes they keep making us wait

(ps feel free to follow me if ur a cutie)

[iyda @notiyda]

GPT 6 Astra Benchmark results GPT 6 Astra might just be AGI?? this model looks insane... what

ARC-AGI-3
• GPT-6 Astra: 98.6%
• Claude Opus 5: 30.2%
• GPT-5.6 Sol: 7.8%

FrontierMath Tier 4
• GPT-6 Astra: 97.6%
• Claude Fable 5.1 / 5: 87.8%
• GPT-5.6 Sol: 83.0%
• Claude Opus 5: 73.2%

DeepSWE v1.1
• GPT-6 Astra: 74.1%
• Gemini 3.8 Flash: 73.7%
• GPT-5.6 Sol: 70.8%
• Claude Fable 5: 69.9%
• Claude Opus 5: 68.8%
• Claude Fable 5.1: 67.4%

ExploitBench
• GPT-6 Astra: 100.0%
• GPT-5.6 Sol: 78.5%
• Claude Opus 5: 70.0%

GPQA Diamond
• GPT-6 Astra: 96.0%
• Gemini 3.8 Flash: 95.3%
• GPT-5.6 Sol: 94.6%
• Claude Fable 5.1: 93.7%
• Claude Opus 5: 93.2%
• Claude Fable 5: 92.6%

SRE-Bench (4 Attempts)
• GPT-6 Astra: 99.2%
• GPT-5.6 Sol: 68.7%

[iyda @notiyda]

GPT-6 Astra is a Mythos level model now and priced the same while achieving nearly 100% on ARC AGI 3

| Evaluation | GPT-6 Astra |
| :--- | :--- |
| ARC-AGI-3 | 98.6% [1] |
| FrontierMath Tier 4 (v2) | 97.6% |
| Agents’ Last Exam | 59.3% |
| AutomationBench | 41.4% |
| BenchCAD | 95.9% |
| DeepSWE v1.1 | 74.1% |
| Terminal-Bench Science 0.1 | 64.6% |
| GPQA Diamond | 96.0% |
| GeneBench Pro | 39.0% |
| MedChemBench (internal) | 49.7% |
| HealthBench Professional (length-adjusted) | 63.4% |
| ExploitBench | 100.0% |
| SRE-Bench (four attempts) | 99.2% |
| Auto-review circumvention (internal; lower is better) | 0% |

[iyda @notiyda]

OpenAI reports Astra scores 97.6% on FrontierMath Tier 4 v2, 74.1% on DeepSWE v1.1, 95.9% on BenchCAD, 96% on GPQA Diamond and 100% on ExploitBench. It also reports a 98.6% score on ARC-AGI-3.

GPT 6 Astra is 98.6% on ARC AGI 3??????? what... this model is going to be insane

[iyda @notiyda]

GPT 6 Astra is considered a "generational leap" in AI

And has improvements in OSWorld 2.0 from 65.7% > 72.6% at 47% less time for the task :)

[iyda @notiyda]

OpenAI is currently edging us and just posting blog posts instead of releasing the GPT 6 Astra it must be super dangerous O_o

take a read though for it it's kind of interesting vv

https://t.co/XjvFBpoLcB

[iyda @notiyda]

GPT 6 Astra is now released and should be fully out in the next 5 minutes

Can't wait very excited to see the progression and improvements from GPT 6 Astra compared to Claude Fable 5.1 and GPT 5.6 SOL

AND we are getting a new Image Generation model and Ultrafast

credit to below :p

[iyda @notiyda]

Guys GPT 6 Astra might have ultrafast it's now being seen inside of the ChatGPT app and some people can select it O_o

[iyda @notiyda]

this is so sad guys we are forced to wait another 15 minutes to see if GPT 6 Astra releases hopefully we get it this time and don't need to wait another 1 hour for the new OpenAI model that will totally mog Claude Fable 5.1

I really think that the new GPT 6 Astra will be a ... HUGE improvement over the current GPT 5.6 SOL model and should be better at doing Frontend Design skill and be better at creative workflows

[iyda @notiyda]

guys... what if we like didn't even get GPT 6 Astra today..? what if it was all a scam what if we weren't getting it until tomorrow...? or worse next week... imagine?

What AI model are you guys running? do you guys activly use the new Fable 5.1 or are you more of a SOL? or even Grok 4.6?

[iyda @notiyda]

GPT Astra in 9 more minutes please????

Welcome GPT 6 Astra

[iyda @notiyda]

Dam bro ChatGPT Astra model is TOO powerful guys it took down every other provider and even took it's self down and blew up

GPT Astra just hacked and shut down Claude,Grok,Cursor and eveni t's self (lol)

[iyda @notiyda]

@scaling01 Hello GPT 6 Astra

[iyda @notiyda]

I can't lie Claude Fable 5.1 has really impressed me with it's fundamental understanding and capabilities on just doing things on it's own rather than needing me to do it for it including PLAYING games for me and testing things inside of a game by taking control it's self even taking screenshots through steam for me and analyzing it honestly quite impressive... some things Claude Fable has been doing is something i'd normally have to do myself now i can just lay in bed and it just does the testing fully end to end for me.

GPT 6 Astra will really need to be on par or better for it to impress me especially considering we all know Astra is most likely shit at frontend as all GPT models are unless they fixed it i really hope so...

Unfortunately Claude Fable 5.1 usage limits are decent but definitely not unlimited the 5 hour limit seems quite harsh on it compared to the weekly limit been hitting the 5 hour limit a lot more on Fable 5.1 compared to before with Fable 5 or Opus 5 hopefully Anthropic give us a little boost or something :P @bcherny

But honestly... really good fucking model some insane capabilities and it just does things properly it just KNOWS it's insane

[iyda @notiyda]

Wait Google ACTUALLY make a model finally????

Gemini 3.8 Flash now has 73.7% on DeepSWE surpassing GPT 5.6 SOL ???