Nahoc (@_nahoc) header
Nahoc (@_nahoc) profile picture

Staff Software Engineer @BoundlessHQ Previously @RISCZero @avalabs

Joined August 2012
Photo by Nahoc (@_nahoc)Photo by Nahoc (@_nahoc)Photo by Nahoc (@_nahoc)Photo by Nahoc (@_nahoc)Photo by Nahoc (@_nahoc)Photo by Nahoc (@_nahoc)Photo by Nahoc (@_nahoc)Photo by Nahoc (@_nahoc)Photo by Nahoc (@_nahoc)Photo by Nahoc (@_nahoc)
The GOAT @emilianobonassi reached out to me earlier this week about agents paying for their open models on Boundless - and he had it up and running within less than a day. See his demo below on Nanocode! Anyone else want to sell Inference, GPU and Databases to Agents?
API keys are for humans. Agents have wallets. @BoundlessHQ now accepts pay-per-token streaming payments in stablecoins via @mpp on @tempo Open a channel → pay as tokens stream. No account, no balance to top up 🧵
3
5
1
37
3,461
Nahoc retweeted
Per-token pricing works for agents because of @MPP sessions. The agent deposits funds and as tokens stream it signs off on the running total offchain. Boundless settles the final amount on Tempo and any unspent balance is returned. Agents can use sessions on @BoundlessHQ now:
Boundless just became the first inference provider on @tempo to accept MPP directly. Give your agent a wallet, and it can buy its own inference, no account or API key required.
3
8
3
39
5,901
Nahoc retweeted
Boundless just became the first inference provider on @tempo to accept MPP directly. Give your agent a wallet, and it can buy its own inference, no account or API key required.
9
11
5
41
8,431
I just built something in under a minute that would've normally taken me hours. On @tradebons, I created an AI agent to: - Track @saylor’s BTC buys - Watch Polymarket odds of BTC dropping to $75K - Alert me if odds cross 57% in a day All I did was describe what I wanted. No coding. No checking multiple platforms. It uses Boundless AI underneath. Bons is live but invite-only: bonbons.app Drop comment for an invite code.
🤝 Paid partnership
24
16
54
5,767
Nahoc retweeted
Meet Bonbons: AI agents that act on your trading ideas for you, watching the market around the clock so you don’t have to.
26
7
6
54
5,843
Kimi K3.1 Pro is now live Day 0 on Boundless. It's being called the most intelligent open model ever released. Built for frontier long-horizon agentic coding. → 1M-token context window. → $1.70 in → $0.20 cached → $8.50 out Try it now with ZDR ↓ inference.boundless.network/…
5
5
1
25
4,694
Nahoc retweeted
testing our crypto trading agent builder on iphone duo lookin good here
2
2
1
18
2,308
Nahoc retweeted
Not everyone wants to build their own trading bot and pay for 20 different services to make it work. Been cooking something with @_nahoc where you just describe your trade idea in plain English, and in seconds you have an always-on agent trading on it. Coming very soon.
8
3
3
19
3,404
Nahoc retweeted
mr incredible learns the truth about artificial analysis intelligence index v5
70
274
30
4,669
154,439
member of the technical debt staff
40
192
8
2,515
61,079
If someone posts a video and prefaces it with, “watch until the end,” skip it immediately.
134
211
19
4,965
57,715
DeepSeek V4.1 Flash 1017 Abliterated is now available on Boundless. → $0.24 input → $0.006 cached input → $0.96 output Arm your defenses. Apply below ↓
1
4
5
20
7,279
________ | We’re [still] hiring | |________| \ (•◡•) / \ / —— | | |_ |_ Open roles: → Applied AI/ML Engineer → Senior Infrastructure Engineer - GPU Compute → Founding Product Manager → Technical Business Developer
If you want to rewrite how the world serves AI, our team is hiring aggressively across BD, Product, and Engineering. Apply now ↓
13
9
1
145
9,809
Introducing Boundless Defensive Access. The security industry is facing an unprecedented volume of attacks. It’s time to arm defenders. Access abliterated models for authorized security testing, starting with DeepSeek V4.1 Flash 1017. Apply: boundless.network/defensive-…
8
7
7
47
17,021
DeepSeek V4.1 Flash 1017 is now available on Boundless. Ranked #1 for TPS on Vercel AI Gateway at 349.90 tokens/sec. Higher throughput means the model generates your response faster. Spend less time waiting ↓ inference.boundless.network/
5
5
5
24
5,836
Opus 5.5 usage check Last week: felt unlimited. Today: 100% of a Max (5x) session gone in 33 minutes. Now locked out for 4 hr 27 min. Anyone else?
101
23
4
654
32,591
Boundless is now available in @OpenCode. Access leading open models with fast, reliable, low-cost inference, right in your coding workflow. Connect your Boundless API key, pick a model, and start building.
15
7
2
35
7,844
Nahoc retweeted
We’re among the best on @vercel AI Gateway for GLM-5.3 Flash throughput and latency + other models. All built by a 4-person engineering team and more optimizations landing soon
ppl are starting to wake up to this 🫣 we (@BoundlessHQ) might serve a few models but, hey we are crunching on throughput, latencies and reliability 🔥 we are in the top 3 😎 infra for inference made it right 🏗️ (data from @vercel AI gateway) 🧵
3
1
1
17
1,138
ppl are starting to wake up to this 🫣 we (@BoundlessHQ) might serve a few models but, hey we are crunching on throughput, latencies and reliability 🔥 we are in the top 3 😎 infra for inference made it right 🏗️ (data from @vercel AI gateway) 🧵
6
4
5
49
8,393
______ | We’re hiring | |______| \ (•◡•) / \ / —— | | |_ |_ Open roles: → Applied AI/ML Engineer → Senior Infrastructure Engineer - GPU Compute → Founding Product Manager → Technical Business Developer
23
22
1
206
20,452