Julian Dropsit

𝐓𝐡𝐞 𝐦𝐚𝐬𝐬 𝐚𝐝𝐨𝐩𝐭𝐢𝐨𝐧 𝐨𝐟 𝐰𝐞𝐛𝟑 𝐠𝐚𝐦𝐢𝐧𝐠 With the milestone of 3 million downloads and 250 million battles played for #HeroesOfMavia @maviagame, I find it timely to discuss the mass adoption of #Web3Gaming. Note: This opinion is purely personal. A (𝑙𝑜𝑛𝑔) thread🧵👇1/6

Second only to Fable? Wait what

@thsottiaux Not having a Canadian accent when using the French live voice would be nice.

How can we make ChatGPT better. It’s already on your phone, but what do you wish would happen more or less of when you open the app. How can it bring you more joy?

@PlayStationEU Funny how there’s no blood in this trailer, while the gameplay footage is full of it. Has anyone else noticed?

Marvel's Wolverine unleashes 15th September on PS5: play.st/4wLHRZB

@jun_song It is true tho that companies benchmaxx their models, that's why I'm always waiting for users' benchmarks / feedbacks on real use cases.

These people don’t even try to put single prompt to test.

90% chance OpenAI resets everyone's Codex weekly limit today 😅

@hrudolph @RexHarrisonsHat @openclaw @steipete request sent, can't dm you on discord until you accept it. 👍

@juliandropsit @RexHarrisonsHat @openclaw @steipete Just dm me. User id hrudolph

OpenClaw v2026.7.1 is out, with 3,063 contributions from 532 contributors! • Web UI and onboarding overhauls • Major iOS, Android and MacOS app work • Telegram, Slack, Discord, and iMessage improvements • GPT-5.6, Muse Spark 1.1, and MORE docs.openclaw.ai/releases/2026.…

@hrudolph @RexHarrisonsHat @openclaw @steipete Can you send me a link to it and lmk which thread I should post in? Thanks!

@juliandropsit @RexHarrisonsHat @openclaw @steipete Can you please hop on discord and help by providing is more details?

@steipete @hrudolph @RexHarrisonsHat @openclaw Channel: Telegram Desktop v6.9.3 x64 OS: Windows 11 OpenClaw v2026.7.1 (issue since 7.1.beta) running on Ubuntu Model: GPT-5.6 SOL xhigh Fast Mode: OFF Text Verbosity: Low Previous working setup: OpenClaw 2026.6.x + GPT-5.5

@hrudolph @juliandropsit @RexHarrisonsHat @openclaw Very curious. I saw this in early May with GPT models and it's fixed. Curious about your setup!

@RexHarrisonsHat @hrudolph @openclaw Yes, it's still happening half of the time on my side. I have to send "do it now" a couple of time to make it work. It makes openclaw almost unusable with telegram at this point. Please fix it @steipete

@juliandropsit @hrudolph @openclaw Been struggling with this very thing today and came out here to see if I was taking crazy pills. A workflow that has been working for months over Telegram just stopped working after this update. The agent kept apologizing but after each reply just stopped working.

@JulienChieze Les keycards de nintendo ne doivent pas être compté comme jeu physique.

SONDAGE ⚠️ Pour les joueurs PlayStation, suite aux récentes annonces, si XBOX et Nintendo continuaient à proposer des jeux en format physique, que feriez-vous ?

GPT 5.6-Sol when I ask it to optimize something

@hrudolph @openclaw Still we're at it, there's a regression we lost the buttons "Allow" and "Deny" on skill approval. They're not showing up anymore

@hrudolph @openclaw Update from the battlefield: it’s actually still happening, but less.

Okay, it seems to work with new sessions now, which wasn’t the case with 2 beta versions I had tried. To me, it still looks like a bug if users have to create a new session after updating OpenClaw. Maybe there should be a message telling us to start a new session after an update. Thanks for your time

@juliandropsit @openclaw Even with new sessions?

In a Telegram group (desktop v6.9.3 x64), whenever I ask my main agent to perform a task using OpenClaw v2026.7.1 (which runs on Ubuntu), regardless of the thread, it keeps replying in a way that suggests it is working on it, but then nothing actually happens. Then if I ask what is going on, it replies with things like: "Sorry, I’m going to do it now", "Right, I’ll do it now—no more false promises" or "You’re right to remind me again" etc. This happens also when he spawns sub agents and tell me he's going to report back once the handshake is done, he now never does. This lazyless happens 25-30% of the time and I have to remind him 2-3 times before he actually starts working. I’m using gpt 5.6 SOL xhigh, fast mode OFF, and text verbosity set to low. I didn’t have this issue with the previous version of OpenClaw and gpt 5.5. Thanks!

@juliandropsit @openclaw Would you be able to provide more info? Would like to address any lingering stability issues and will do everything we can to do so! Sorry about that.

@jun_song I'm really disappointed with SOL tbh. Benchmarks say it's smart but real-world use cases feel otherwise, since day 1 btw.

How to fix a slow AI model : Remove its brain and make it stupid.

@baguley61 @Spshulem With a $200/month plan, it would take 16 years to reach $40k.

@Spshulem You aren’t doing anything with any $20 a month plan. Except testing the models out.

“So you spent $40K to run GLM 5.2 locally?” “Yes, Dave” “And you did all this to save $20 a month on Cursor” “Thats right, Dave”

@OpenAI @dr_l_alexandre

GPT-5.6 is a major step forward for health intelligence. Across the lineup, we’re delivering stronger performance at lower cost: GPT-5.6 Luna outperforms GPT-5.5 at its highest reasoning setting while costing 25x less. Together, these advances raise quality while making advanced models accessible to more people globally.

@DavidOndrej1 Explain.

this is worse than i expected