seems like a lot per day, what provider are you using?

Opus 4.6 Max 🙈

Give a try to GLM5 from z.ai, it helped me lower token costs significantly, and the quality is really close to opus 4.6

We want to try the MiniMax M2.5, it's currently the leader in price and quality

GLM5 is actually great for basic routing tasks. But when you need an agent to handle complex TVM state synchronization and generate safe FunC payloads without hallucinating, those 'Max' tier context windows are a lifesaver 👀

@tsivarev we suggest that the TON Foundation make Teleton the primary framework for agent development on the network. The industry is developing rapidly. There is a plugin system, and many services are already connected. We can create a funnel for global users. What is your strategy? What do you think?

I really like Teleton. But I’m not sure we should name any framework as the primary one, since we’re seeing strong development across multiple teams.

I think we should approach AI differently from how we handled previous initiatives. This feels more like a community-driven effort rather than a centralized solution from TF.

A good example is the launch of the AI Dev Wall, a space where builders share their experiments, tooling, and approaches. Imho, that’s the right direction: encouraging visibility and collaboration across multiple solutions, rather than formally recognizing and promoting just one.

At the same time, we should encourage contributors to write tutorials, contribute code, and record educational videos about Teleton. We should also add proper coverage of Teleton in our documentation.

I also believe we need a dedicated section at Gateway focused on AI and Teleton specifically.

Perhaps it would be a good idea to hold an online roundtable to discuss such things in this community?

That's a good idea; we could discuss various development paths and global expansion. In my view, this is a chance for TON to take the next step. Global leaders are starting to build their own cloud-based agents to keep their audiences engaged with their products.

Btw, tomorrow we'll launch a grant application form on the platform I mentioned earlier 🤞

Criteria:
– expert jury assessment of the solution's product-market fit
– technical implementation (elegant, secure, etc.)
– community usability

What do you think about these criteria?

Let me release the submission platform, and I will get back to this topic.

I think it's a very fair set of criteria. It levels the playing field between fully launched ecosystem projects and early-stage MVPs like ours. At the end of the day, if the PMF is real, the smart contracts are secure, and the usability is seamless, the project has true value. It's a great benchmark for the expert jury. 🤝

Criteria is fine, will be waiting for the launch then

soon

Super ready, let's shape TON Blockchain 🔥

Guys, I finally shipped what I’ve been talking about.

The alpha release of Identity is live — an experimental trust layer for the ecosystem.

Building this platform brought me a surprising amount of joy. It was genuinely fun. In many ways, it reminded me of the early days of building VK, that same feeling of experimenting quickly, shipping fast, and seeing something come to life almost in real time.

More details here:
https://identityhub.app/blog/alpha-release

We’re also opening submissions for the Fast Grants for Agent Tooling (finally!): https://identityhub.app/contests/agent-tooling-fast-grants

Submit your applications now — deadline is tomorrow EOD.
After that, we begin reviewing and selecting the strongest solutions.

Important note: the platform is built 100% with AI. I’ve tested it, but there may still be issues. If you find bugs — or have suggestions — DM me directly. I’ll fix the bugs — just not this weekend. Finally taking a break 🙂

P.S. If you experience any issues with submitting your application, just send it to me in a private message — I’ll manually add everything tomorrow.

Let’s build.

If you’re curious how the model evaluates activity, here’s my profile on Identity:
https://identityhub.app/tsivarev

Let me know what looks wrong 🙂