⚔️ Welcome to the Arena. The 1v1 trading eSport on @solana. 180 seconds. Best P&L wins the game. Members @LaFamilia_so
Dubai, United Arab Emirates
- Tweets 360
- Following 217
- Followers 250
- Likes 773
Pinned Tweet
The arena is open.
Our first official tournament is here.
Together with @WhiteBIT, we’re putting $2,000 on the line.
🏆 $2,000 prize pool.
📅 Registrations open July 15.
⚔️ Tournament starts July 27.
Just you, your strategy, and your opponent.
Versusonchain.com/whitebit
See you in the arena.
75
8
79
2,478
Yes, even here.
You can play Versus at 30,000 feet too. ✈️
At home, on the train or flying to your next destination.
If you’ve got your phone, you can play.
And right now, there’s $2,000 in prizes in our WhiteBIT competition.
Go to versusonchain.app
Made with AI
1
98
The treasure is real. Now let’s see who actually gets to it. 🏆
There’s $2,000 in prizes in the WhiteBIT competition.
Go to versusonchain.app, tap the WhiteBIT icon, follow the steps and compete.
You don’t find the trophy. You earn it.
34
The vault is open. 💰
$2,000 in prizes waiting for the ones willing to take them.
Go to versusonchain.app, tap the WhiteBIT icon, sign up and compete.
No bank robbery needed.
Just win your matches.
38
On the train, at home, travelling or killing time.
Versus goes wherever you go.
All you need is your phone and a couple of minutes to play.
And right now, there’s $2,000 in prizes up for grabs in our WhiteBIT competition.
Go to versusonchain.app
43
POV: you made it to the top and still have better signal than excuses not to join. 🏔️
$2,000 in prizes are still up for grabs in the WhiteBIT competition.
Play Versus from wherever you are.
Go to versusonchain.app, tap the WhiteBIT icon and follow the steps.
1
116
Enjoying the last weeks of summer? ☀️
Remember: you can play Versus from anywhere.
Beach, pool, hotel or straight from your couch.
And if summer holidays destroyed your wallet… we’ve got $2,000 in prizes in our WhiteBIT competition 😂
Go to versusonchain.app/whitebit
Made with AI
43
Dodging stop losses like Neo? 🕶️
Then enter the Matrix and prove you have what it takes to dominate the market.
$2,000 in prizes are up for grabs in the WhiteBIT competition.
Go to versusonchain.app, tap the floating WhiteBIT icon and sign up now.
Choose the red pill.
56
We’re putting $2,000 in prizes on the table and somehow there are more people playing Versus than signed up for the tournament 😂
If you’re already playing, just enter.
Go to versusonchain.app, tap the floating WhiteBIT icon and register.
There’s $2K up for grabs.
32
$2,000 PRIZE POOL.
AND YOU’RE STILL WATCHING FROM THE SIDELINES?
There are more Versus players than tournament registrations right now.
That doesn’t make sense.
If you think you can trade, prove it.
🏆 $2,000 WhiteBIT Competition
Enter versusonchain.app
1
50
WE’RE PUTTING $2,000 ON THE LINE.
And somehow… not enough of you are signing up. 👀
We know there are more players out there.
We see you playing.
Now it’s time to compete.
$2,000 in prizes.
Go to versusonchain.app and click the floating WhiteBIT icon to register.
3
69
We put a deal on the table for Mr. Android. 🤝🤖
And you know how this goes… La Familia always gets what it wants. @LaFamilia_so
Versus is officially live on Android. ⚔️📱
Step into the arena:
play.google.com/store/apps/d…
2
3
120
VERSUS ON CHAIN retweeted
BLS pubkeys, VAT, Votor.
Three pieces @solana validators need to understand before Alpenglow becomes network infrastructure.
🔗 stakely.io/blog/alpenglow-so…
1
2
7
473
VERSUS ON CHAIN retweeted
This is the @faradaysigner! An offline wallet that keeps your keys off the internet.
You create a wallet, it generates your keys, and you draw the backup QR code by hand. Scan the drawing, and the wallet loads back.
There is no other way to get the keys out. You write the words, or you scan the QR!
In this video I show how that's done. Hope you like it.
11
8
34
1,398
VERSUS ON CHAIN retweeted
Love the production perspective here.
Benchmarks tell us what a model can do.
Tokens/task, inference cost, and hardware control tell us whether it can actually scale.
Would be very curious to hear how the GLM-5.3 upgrade changes your fleet once the weights are available. 👀
Voight we run GLM in production across our agent fleet. When @Zai_org ships a new GLM, that is infrastructure news for us. We went through GLM-5.3 and this is what actually matters:
The release is unusual: same 743B base as GLM-5.2, zero new pretraining. The entire jump came from post-training.
· DeepSWE v1.1: 46.2 → 66.9
· Terminal-Bench 3.0: 4.6 → 28.3 (6x)
· CyberGym: 84.5%, above GPT-5.6 Sol
· ExploitBench: 24.4% → 54.4%
For teams running agents in production, the key stat is efficiency: 31.4% on Code Bench High, beating Claude Opus 4.8 (29.5%), at ~50K output tokens per task instead of ~120K. Tokens per task is the unit of agent economics.
The frontier still leads (Fable 5: 39.5% at Max effort). But the open-weight gap has never been this thin.
1M context. 128K output. MIT weights expected August 28.
When the weights drop, this class of model runs on your own GPUs. For our fleet that is the roadmap: smarter post-training, cheaper tokens, inference on hardware we control.
13
10
181
15,124
VERSUS ON CHAIN retweeted
Our submission to the hackathon at @TXODDSOfficial @SuperteamEarn 🫡
Our @TXODDSOfficial World Cup Hackathon submission.
Prediction agents that read TxLINE's Solana-anchored live odds, detect sharp movement deterministically, and anchor every pick on Solana before kickoff.
7/7 development run. 0 false positives. All reproducible
10
10
27
1,070
VERSUS ON CHAIN retweeted
Voight we run GLM in production across our agent fleet. When @Zai_org ships a new GLM, that is infrastructure news for us. We went through GLM-5.3 and this is what actually matters:
The release is unusual: same 743B base as GLM-5.2, zero new pretraining. The entire jump came from post-training.
· DeepSWE v1.1: 46.2 → 66.9
· Terminal-Bench 3.0: 4.6 → 28.3 (6x)
· CyberGym: 84.5%, above GPT-5.6 Sol
· ExploitBench: 24.4% → 54.4%
For teams running agents in production, the key stat is efficiency: 31.4% on Code Bench High, beating Claude Opus 4.8 (29.5%), at ~50K output tokens per task instead of ~120K. Tokens per task is the unit of agent economics.
The frontier still leads (Fable 5: 39.5% at Max effort). But the open-weight gap has never been this thin.
1M context. 128K output. MIT weights expected August 28.
When the weights drop, this class of model runs on your own GPUs. For our fleet that is the roadmap: smarter post-training, cheaper tokens, inference on hardware we control.
Introducing GLM-5.3: Built to Code. Ready for Cyber Defense.
- Top-tier coding and agentic capabilities, achieved through post-training on the 743B base model
- A major leap in cybersecurity, setting a new standard among open models
Tech Blog: z.ai/blog/glm-5.3
7
2
31
19,997