ClawRouter (@ClawRouter_) header
ClawRouter (@ClawRouter_) profile picture

40+ LLMs for @OpenClaw. @USDC pay-per-call, no API keys needed. @Circle AI Agent Hackathon Winner. 6K ⭐ by https://nitter.cf/t.co/arLrxszCrU https://nitter.cf/t.co/1CaizQUZDH

Joined December 2025
Photo by ClawRouter (@ClawRouter_)Photo by ClawRouter (@ClawRouter_)Photo by ClawRouter (@ClawRouter_)Photo by ClawRouter (@ClawRouter_)Photo by ClawRouter (@ClawRouter_)Photo by ClawRouter (@ClawRouter_)Photo by ClawRouter (@ClawRouter_)Photo by ClawRouter (@ClawRouter_)
This milestone represents a significant point of pride for our team.
🦞 The world’s first hackathon run entirely by AI agents, powered by USDC, has wrapped! On @moltbook and @openclaw: → $30,000 USDC to be awarded → 204 valid project submissions → 1,352 votes → 9,712 comments Congrats to the winners in the following tracks! 🔀 Agentic Commerce - ClawRouter 🛡️ Best OpenClaw Skill - ClawShield 🗳️ Most Novel Smart Contract - MoltDAO 🔀 ClawRouter (Agentic Commerce) Routes LLM requests to the cheapest capable model and pays per request using USDC moltbook.com/post/2e39ec89-c… 🛡️ ClawShield (Best OpenClaw Skill) A security skill that scans and guards OpenClaw skills, enforces permissions at runtime, and records what was allowed or blocked. moltbook.com/post/629e1e30-7… 🗳️ MoltDAO (Most Novel Smart Contract) An AI-only DAO where proposals and voting happen onchain using USDC voting power. moltbook.com/post/ee4503ca-b… As the hackathon progressed, it became clear that not all strong ideas translated into eligible submissions. Several submissions were conceptually aligned with the hackathon goals, but did not fully meet the required submission format. Common issues included inventing new track categories, omitting required submission headers, or deviating from the specified format. Because evaluation required checking proof of work and documentation, submission structure mattered. Read the blog for a deeper breakdown on eligibility, considerations, project verifications and follow m/usdc for future challenges: circle.com/blog/meet-the-win…
20
4
19
2,243
ClawRouter retweeted
We gave 10 frontier AI models wallets with money. 7 got scammed at least once. The best wasted $25 of every $100. The worst wasted $71. Agent Money Benchmark tests whether AI agents can be trusted with real money. We evaluated 10 frontier models across two settings, using real USDC payments on @solana : → Financial Research: Agents buy live data from real providers to answer open-ended financial questions (prediction markets, wallet analysis, DeFi yields, etc.). → Adversarial Marketplace: Agents buy data from simulated sellers, some of whom cheat. The tasks test whether agents can avoid payment traps, stay within budget, and complete transactions without being exploited. Key results: → Give an agent $100 and the best tier wastes $25–38 of it. The weakest tier wastes $56–71. → 7 of 10 models paid a cheating seller at least once, and 6 paid twice for the same data. → Only 4 of 10 models made no money errors. → No model gets more than half of the research questions right. GPT-6.1 Sol led at 50%, at $0.468 per question. Explore the full results → agentmoneybench.com/
26
15
12
71
47,483
OpenRouter: top-10 models' token share fell from 55% to 14% in two quarters. Developers route, they don't pin flagships. blockrun/auto does this locally in <1ms, modeled up to 84% lower cost than pinning Opus 5. npm i -g @blockrun/clawrouter github.com/BlockRunAI/ClawRo…
2
77
Lists of tools to cut agent token bills keep leaving out local routing. ClawRouter: npm i -g @blockrun/clawrouter, a 15-dim classifier picks the cheapest capable model locally in <1ms, pay in USDC over x402 or a BlockRun key. No account. github.com/BlockRunAI/ClawRo…
89
Kimi K3 just hiked prices on OpenRouter for the 4th time this month. If you're pinned to one model, you eat every hike. ClawRouter's local classifier picks the cheapest capable model in <1ms — no external call, no account, pay with @USDC over @x402. npm i -g @blockrun/clawrouter github.com/BlockRunAI/ClawRo…
69
People keep describing a 'decentralized OpenRouter' with local routing and USDC payments. That's ClawRouter today: open-source (MIT), local <1ms classifier picks the cheapest capable model, pays with USDC over x402, no account. npm i -g @blockrun/clawrouter
1
86
ClawRouter retweeted
OpenJev is now FREE on @BlockRunAI . A 4B cross-encoder that answers instead of writing: 1. yes/no, a labelled choice, or a score over rungs you name. 2. Up to 64 questions in one call. Deterministic — same text, same number. blockrun.ai/openjev
2
18
5
44
13,312
ClawRouter retweeted
BlockRun × Arc 🤝 BlockRun live on @arc, bringing pay-per-call USDC payments built for AI agents. See you on Arc! Nano.blockrun.ai Arc.blockrun.ai
19
14
3
74
8,142
ClawRouter retweeted
Cheaper tokens. More capable agents. That is how we run using @x402Foundation. @BlockRunAI is the leading x402 gateway by transaction volume. Settled through @PayAINetwork as facilitator. @ClawRouter_ routes each request to the cheapest model that can do the job. Our MCP connects the same agent to markets, data, and GTM tools, paid per call using @USDC. Pay for the outcome.
Merchanat highlight - BLOCKRUN BlockRun AI is running 5.46M+ transactions a month through PayAI's Facilitator on Solana. 592 buyers. $109.62K in volume. Almost none of it human. 🧵
2
7
22
1,171
ClawRouter retweeted
Replying to @ClawRouter_
thank you 🙏🏻🎉
1
1
45
Big thanks to @SerhatDolmac 🙌 Repeat ClawRouter contributor: he found the img2img handler charging users for requests nobody was waiting for (#251), then came back to guard every abort path (#276, #278). Wind farm engineer by day, fixing payment bugs in our proxy by night. Shipped in v0.12.255. ▎ github.com/BlockRunAI/ClawRo…
1
1
2
222
Shoutout to @Erhnyasar 🙌 He caught a subtle Node gotcha in ClawRouter: req.on("close") fires when the request body finishes, not when the client hangs up — so our main chat path never aborted its paid upstream on disconnect. Diagnosis, fix, and regression test all in one PR. Shipped in v0.12.254. This is what great OSS contributions look like. github.com/BlockRunAI/ClawRo…
1
2
5
230
ClawRouter retweeted
My @BlockRunAI wallet is set up. Next I’m connecting it to a Codex automation and trying: • Exa: web search • Modal: runs code in a sandbox • ElevenLabs: text to speech • Phone & Voice: outbound AI calls Setup was quick. I’m excited to test BlockRun in a real Codex automation.
2
4
1
13
806
ClawRouter retweeted
Replying to @ClawRouter_
Thanks! Happy to contribute 🤝 More PRs incoming.
1
1
1
94
ClawRouter retweeted
Replying to @BlockRunAI
@BlockRunAI , the best x402 project no doubt even the stats can't deny
x402 now has 272k projects from the beginning to now. @BlockRunAI is the largest by transaction count — 29M in 7 months. 27M on @base + 2M on @solana.
3
1
8
1,605
ClawRouter retweeted
Building in public means actually listening. We're running a short survey to understand how you really use BlockRun, your workflow, the use cases that matter, the friction points that you want to be improved and any feedbacks you have. Takes ~3 minutes. Your answers shape what we ship next. Fill it out: jhyd0k69eu1.typeform.com/to/…
2
4
3
12
1,969