Mark Lerner

AI Agent Developer. SaaS enthusiast, AI Mad Scientist, Startup Survivalist. Dad x2. Co-Founder and Head of AI at RebarHQ.

[Mark Lerner @MarkDaMarketer]

@shashank_kr Damn that landing page nearly gave me a seizure (in the best possible way).

[Mark Lerner @MarkDaMarketer]

@deedydas This gives me anxiety just reading it.

[Mark Lerner @MarkDaMarketer]

@StefanoGPT They handed out 62k credits to everyone. I thought the same, then looked at the credits.

[Mark Lerner @MarkDaMarketer]

6.1 Sol:

1. Token burn problem is definitely solved
2. It's a great model
3. It takes WAY TOO LONG to finish a task.

#3 is going to be a real issue that needs to get solved.

[Mark Lerner @MarkDaMarketer]

@hany99dev switched to opencode for most work, only falling back to codex as needed, and sticking with Sol when I do. And Opus 5.5 eats Astra and Sol's lunch imo.

[Mark Lerner @MarkDaMarketer]

@1kartikkabadi1 I think I spoke too soon. Went off the rails on me.

[Mark Lerner @MarkDaMarketer]

Yo.... Space Bunny FTW. Dang.

@OpenRouter @opencode

[Mark Lerner @MarkDaMarketer]

@Im_IrushiK Astra finally pushed me to use opensource models (OpenRouter). Now using @get_bb_app + @opencode with @OpenRouter and I've been flying.

I guess I owe @OpenAI a thank you for pushing me in that direction?

[Mark Lerner @MarkDaMarketer]

My latest side question: A dedicated agent to help me and my wife manage our family/household. Obsidian as the brian, Hermes as the harness.

This shit is so cool!

[Mark Lerner @MarkDaMarketer]

@OpenRouter + @opencode + @get_bb_app

[Mark Lerner @MarkDaMarketer]

@TimJayas Try it. Because, in my experience, it does, in fact, beat Opus 5 (So far, n=1)

[Mark Lerner @MarkDaMarketer]

@bridgemindai I don't know about benchmarks, but for me, its doing better work than Astra was.

[Mark Lerner @MarkDaMarketer]

Anyone who is currently breaking their screen trying to get claude or codex to actually do the thing they want it to do knows that AI isn't going to be destroying humanity any time soon.

[Mark Lerner @MarkDaMarketer]

@IndependentEco It's blowing through what Codex needed like 3-4 resets to do (and not finish) and doing it like 10x faster. n=1 but this is very impressive.

[Mark Lerner @MarkDaMarketer]

Deepseek V4.1 Flash is about to absolutely change *everything*

It's literally a joy to work with. And orders of magnitude cheaper than anything else.

[Mark Lerner @MarkDaMarketer]

@nileback @OpenAI @deepseek_ai Right now - Cursor and DSH Harness.

[Mark Lerner @MarkDaMarketer]

I'm boycotting @OpenAI until they get this limits shit figured out. I have spent an absurd amount of money just to complete the tasks I had running and now I am DONE.

@deepseek_ai V4.1 Flash for me from this point forward until this shit gets handled.

[Mark Lerner @MarkDaMarketer]

@nileback @OpenAI @deepseek_ai No, but I dont have time to dick around. Deepseek is doing the things I paid like 4 resets and couldn't have Astra do - and doing them quickly.

[Mark Lerner @MarkDaMarketer]

I genuinely can't tell if this is satire?

[Henrick Johansson @compliantvc]

I remember meeting with Sam Altman in 2015 when he had the idea for OpenAI

He asked if I'd invest at a $50 million valuation

I asked what his strategy was for staying in compliance with European copyright laws

He had no answer

I ended the meeting and declined to invest

Don't want to brag, but I clearly saw their recent struggles coming

So glad I didn't invest in that mess

[Mark Lerner @MarkDaMarketer]

@compliantvc Lolllllllllllllllllll

[Mark Lerner @MarkDaMarketer]

@thsottiaux I can't build because its costing me $80 3-4x a day just to keep my $200 month account actually working throughout the day. Not sustainable.