Gemini 4 Flash seems to be surfed in Devin CLI
I found a new model family called Hyperion in Devin CLI v3000.12.17
It has 5 variants: hyperion, hyperion-minimal, hyperion-low, hyperion-medium, hyperion-high
Hyperion appears between Gemini 3.5 Flash Lite and Gemini 4 Argon in the model list
This makes me think Hyperion could be Gemini 4 Flash although it's still unconfirmed
Show more
Good News for ChatGPT Plus users, OpenAI is considering introducing a $50 Pro plan
This is exactly what many Plus users have been asking for, Plus limits can sometimes feel restrictive for heavy users but many college students, developers and everyday users simply can't afford $100/month for Pro
A $50 middle tier with higher limits and access to more futures would be a huge upgrade
Hopefully OpenAI makes it happen ASAP
Show more
Would a Pro 50 make sense? Wdyt?
Anthropic may be preparing a new Claude model family, potentially codenamed "Silicon"
Following the "claude-donut-eap" discovery from earlier we've now spotted a mysterious model family called "Silicon"
Silicon has five reasoning variants: Low, Medium, High, XHigh and Max
it's listed directly after Obsidian 3, while Donut appears in another section alongside Claude Opus 5.5, Mythos 5.1 and other Claude models
Both appeared in the latest Devin CLI build and were absent from the older version
This could point to new experimental models undergoing internal testing at Anthropic with Silicon potentially connected to the Donut
Show more
🚨 Fable 5.5's codename could be `claude-donut-eap`
this new codename `claude-donut-eap` appeared Devin CLI new build today
I haven't seen this codename before it's a new codename added today
We are seeing many signs of fable 5.5 now I wouldn't be surprised if Fable 5.5 is being preparing for a release as early as next week
Show more
New OpenAI model codenames spotted in Devin CLI
Cactus and Kraken are linked to OpenAI and appear between GPT-6 Luna and GPT-6 Astra
Both have 5 reasoning levels: low, medium, high, xhigh and max and They were absent from the older build (v3000.11.239)
What is OpenAI cooking?
Show more
🚨 Fable 5.5's codename could be `claude-donut-eap`
this new codename `claude-donut-eap` appeared Devin CLI new build today
I haven't seen this codename before it's a new codename added today
We are seeing many signs of fable 5.5 now I wouldn't be surprised if Fable 5.5 is being preparing for a release as early as next week
Show more
So apparently, Fable 5.5 isn't dropping until OpenAI releases GPT-6.2 Astra…
And OpenAI isn't releasing GPT-6.2 Astra until Fable 5.5 drops???
New on our Science Blog: How AI is beginning to change scientific modeling.
Scientific models help researchers explain and predict real-world systems—like how a single nerve cell fires, or what happens when neutron stars collide. Recent advances have transformed how scientists fit complex models to data. But scientists still typically write the models manually, and usually need to weigh up several competing versions, each of which might explain the data in a different way.
In a guest post, Jakob Macke, Daniel Gedon, and Stefan Wahl describe how AI is starting to help with these parts of the process. By writing, running, fitting, and revising scientific models as code, AI can explore many possible model structures, and find out which explanations the data support.
Show more
Translation: GPT-6.2 Astra isn't coming today
OpenAi is waiting for Anthropic to release Fable 5.5 first so OpenAI can release gpt 6.2 astra and interrupt its IPO
A new way to measure how AI agents handle the routine analyses of wet-lab science—BioBasicsBench.
Developed with Ligo, a biotech company focused on enzyme design, the benchmark draws on 100 tasks from real experiments to test whether agents can interpret raw lab data, identify errors, and recognize when the evidence doesn’t support a conclusion.
Accelerating biological research starts with getting the everyday work right. BioBasicsBench shows where models need to improve, and helps us track what those gains mean for scientific progress, biosafety, and biosecurity.
Show more
Gemini 4 Hydrogen – Aldermere Test
Google is testing two checkpoints on Arena This is checkpoint B
It's just as fast as the other checkpoint but the output quality seems lower than checkpoint A
Show more
Qwen4-27B is dropping today!!
Get ready 👀
Let's see what the RSI has done
What? Anthropic now removed Mythos 5.5 from the website
Was it accidentally exposed ahead of an announcement?
Mythos 5.5 is showing up in Anthropic website
A new reference to "Claude Claude Mythos 5.5" has appeared, alongside an image URL hosted on Anthropic's CDN
I think release is imminent maybe next week??
Show more
Mythos 5.5 is showing up in Anthropic website
A new reference to "Claude Claude Mythos 5.5" has appeared, alongside an image URL hosted on Anthropic's CDN
I think release is imminent maybe next week??
Show more
Mythos 5.5 is already listed HTML source code of Anthropic's official website
Thanks to the anonymous person who sent me the tip via email :)
GPT-6.2 Astra is showing up on codex
The UI still GPT-6 Astra but the API response breakdown now shows "gpt-6.2-astra"
I think OpenAI is testing a new Astra version
🚨 Gemini 4 Carbon: 3D Car Racing Game 🏎️
Carbon one shotted in just 1 prompt an entire playable car racing game generating everything from scratch in around 20 minutes
this is the new Carbon 4 checkpoint
Show more
ChatGPT users, I have another bad news for you
OpenAI has introduced device limits for paid accounts using Codex One device can only be used with a limited number of paid accounts (Go, Plus, Pro 100/200/500) signing out of another account won't free up a slot
Business, Enterprise, Edu, and Free accounts aren't affected by this limit
If you need more Codex usage OpenAI points users toward purchasing additional credits through Usage & Billing
Show more
This Bad news for ChatGPT Plus users
We already had pretty limited usage to begin with and now OpenAI is moving toward a shared usage pool between ChatGPT and Codex
This means your ChatGPT and Codex usage will come from the same allowance, Unless OpenAI increases the overall limits significantly, this could make things even worse for Plus users
We really need more usage, not fewer ways to consume it
Show more
ChatGPT users, OpenAI is merging usage allowances
Starting in December Plus, Pro, and Business plans usage allowance will be shared across ChatGPT and Codex
as plus users this will impact us badly I hope openAi shouldn't do this
Show more
ChatGPT users, OpenAI is merging usage allowances
Starting in December Plus, Pro, and Business plans usage allowance will be shared across ChatGPT and Codex
as plus users this will impact us badly I hope openAi shouldn't do this
Show more
Alibaba is inviting selected developers to test qwen 4 models
"qwen-latest-series-invite" spotted in official Alibaba Qwen Code
meaning selected people may get access before the public release next week
Show more
GPT-6 Astra Is Seriously Underrated
I've been heavily testing GPT-6 Astra on X High today and honestly it's WAY better than most people think The problem is Most people haven't even experienced its full potential because it burns through usage limits so quickly
From my testing it's better than every Fable model and in some areas it even beats Opus 5.5 GPT-6 Astra is basically OpenAI's version of Fable but with an even higher ceiling
And if GPT-6.1 Astra really arrives next week with another significant improvement anthropic will fall behind if they don't release fable 5.5
Show more
Day 26 of Tibo's 28 day challenge
The next two days will be a Banked resets, I'm not sure what to expect, but Tibo might be hinting at something big I hope it will be a new model for the next week
Let's see what OpenAI is cooking
Show more
Day 26/ You can now easily switch between multiple accounts on mobile, and coming soon to desktop too. Next two days will be resets… banked… do with this what you may 👀.
Show more
GPT-6.1 Astra will surprise a LOT of people
I think it could make people feel the same way they felt when Opus 5.5 first dropped That model genuinely amazed people and reminded everyone what frontier AI models are capable of
OpenAI's models have been missing a taste and a certain level of intelligence creativity and that extra spark But from with Astra, it looks like OpenAI may have finally figured it out
They might have mastered the missing piece
Show more
