Blake

Closeted AI Nerd ¯\ _(ツ)_/¯ | Denver Sports Supporter | Building something to help cure rare diseases | Your wrapper is not a startup lol

[Blake @BlakeMileHigh]

About 20 minutes away from rolling back to the Operating Room to remove a benign tumor in my brain.

Bummed that I won’t be fully “with it” for Sunday’s game, but I know that @CodyRoarkNFL, @NickFerguson_25, @SayreBedinger, and @BrandonPerna have me on the recap.

Cya on the other side, folks 🫡💯

[Blake @BlakeMileHigh]

@thsottiaux Waaaaait it wasn’t a third party harness that most people had never heard of? Weird 🙄!

[Blake @BlakeMileHigh]

@thsottiaux It’s becoming increasingly apparent just how much resets have become a crutch for disregarding genuine concerns regarding drainage rates. Bread and circuses lol.

(I complain, but I will absolutely be using the reset. I am a hypocrite.)

[Blake @BlakeMileHigh]

@thsottiaux Tibo this is such an intentional misconstruing of facts. You’re better than this. There’s a difference between usage limits being lowered and usage draining quicker. Your post is intentionally misleading at best dude.

[Blake @BlakeMileHigh]

@thsottiaux Tell me about it. Really glad I escalated a ticket to support just to hear nothing back and then being forced to ration my usage for the last few days LOL.

[Blake @BlakeMileHigh]

@thsottiaux I’m jealous—it must be nice to have sufficient usage to enable you to ship after midnight. Seriously though, what is up with limits Tibo? This is not normal.

[Blake @BlakeMileHigh]

@RealFlyingHigh @thsottiaux I'm using a custom orchestrator, but if you ask specifically for a script to monitor/log telemetry then it has access to the data (I don't know if it's logged elsewhere locally by default).

[Blake @BlakeMileHigh]

@thsottiaux Live look at me after spending 76% of Pro plan usage in 90 minutes bc I thought the reset was going to happen at any moment

[Blake @BlakeMileHigh]

@thsottiaux After blowing through 60% usage on a “security incident” that ended up being a mirage due to human miscommunication, you’re my hero rn

[Blake @BlakeMileHigh]

@thsottiaux Everyone who still has that 07/31 reset banked:

[Blake @BlakeMileHigh]

@thsottiaux I mean, it literally saved my life by identifying that I possess a novel, harmful genetic mutation. And I’m not saying that that’s pretty much as impactful as it gets, buuuuuuut…

[Blake @BlakeMileHigh]

@thsottiaux For context, I mean files like this. They don’t actually open.

[Blake @BlakeMileHigh]

@thsottiaux When an agent creates an artifact running through Codex desktop, an error occurs when users try to access the artifact on their phones since it’s not technically on the device. If you could let us open them, that’d be great

[Blake @BlakeMileHigh]

@thsottiaux You don’t know it yet, but you just did an amazing thing for people living with rare and undiagnosed diseases, Tibo. Give it 6 months and I promise you’ll see when the Harvard/Stanford paper is published lol. Which sounds like a grift, but I am serious. Thanks g

[Blake @BlakeMileHigh]

@crusoeselkirk It’s wild how you managed to type all of those words and still not make a single point lol. But yes, Robin, keep shilling for Dario. Everyone really, really cares and wants to hear your opinion, I promise.

[Blake @BlakeMileHigh]

@thsottiaux Hey @thsottiaux as nice as an approximate rollout time would be, confirming that we're not going to get got by some "Actually it's rolling out to government sanctioned accounts first, then enterprise accounts then..." bs & whether Sol will be in Codex day 1, would go a long way.

[Blake @BlakeMileHigh]

Google fumbling Gemini so severely still blows my mind. Despite arguably having the most raw resources at their disposal:

❌ No model remotely approaching GPT-5.5 or Opus 4.8 or Grok 4.5, much less Sol/Mythos or even Fable/Terra
❌ No agentic AI development platform
❌ Refusal to actually parse long-form documents; truncates them even when asked not to
❌ No skill creation or use
❌ Last competitive model released in Feb ‘26
❌ No full functionality desktop app
❌ Did I mention that there’s no agentic AI development platform?
❌ Not that the lack of agentic AI matters considering Gemini has given me zero reason to believe it possesses any advanced programming capabilities, anyway lol
❌ Lack of updates has resulted in a feeling of it being antiquated and something I only use for fact-checking straightforward concepts/info/news
❌ Isn’t even the best at making/designing Google Docs/Sheets

✅ But at least we have nano banana v972 (tbf it’s actually decent)
✅ Music creation is also neat
✅ At least it isn’t run by a doomer megalomaniac who wants to consolidate power like Anthropic

After GPT-5.2, OpenAI realized they needed to stop trying to be cute and just focus on training and making more advanced models. GPT-5.4 was good. GPT-5.5 was great. And today, GPT-5.6 Sol is dropping.

The only way Google saves Gemini is to stop trying to put out sideshow projects while they’re multiple generations behind lmao.

That’s the only way. Not “the only way to take the top spot” (that ship has sailed lol), more like “the only way to claw back to literally any relevance at all.”

You’re welcome, Josh.

[Blake @BlakeMileHigh]

@DeryaTR_ @romainhuet This unironically gives me confidence that Sol will be able to tell us what the function of the mitochondria is. Excited to see what you’ve got cooking 👀.

[Blake @BlakeMileHigh]

@Lentils80 Oh they’ll refine the classifiers alright… after they switch to $10/input API calls that is lol. Shameless company ngl

[Blake @BlakeMileHigh]

@DeryaTR_ Dr. Unutmaz, do you know whether GPT-5.6 variants will permit non-institutional users to utilize them for the development of clinical tools? The Fable fiasco left me anxious that future frontier models would inherit the same baseline restrictions that Dario blessed us with.