Cascade

Pay-per-use APIs for AI agents. Inference, Twitter, Reddit, web - one wallet, no API keys. Built on @x402 + @mpp

[Cascade @cascade_fyi]

Most agent APIs force a single payment method. Surf returns all supported options in one response. The agent picks.

No API keys, no accounts. Just an MCP tool call that settles payment before returning data.

9 tools for live social and web surfing - each one pay-per-call.

[Cascade @cascade_fyi]

"My agent can't read half the web" is a real problem and almost nobody talks about it. Native fetch gives you an empty SPA shell and the run quietly degrades.

glim is the fix we've been running in prod - pay per call, no subscription.

[glim.sh @glim_sh]

Agents bounce off half the web: bot walls, SPA shells, JS-heavy pages.

glim_web_fetch reads the pages plain fetch can't. Plus web search, Reddit, GitHub, YouTube. One MCP endpoint, pay per call.

claude mcp add --transport http glim https://t.co/9278Uln4qX

[Cascade @cascade_fyi]

@shafu0x sent dm

[Cascade @cascade_fyi]

Most agent APIs force a single payment method. Surf returns all supported options in one response. The agent picks.

No API keys, no accounts. Just an MCP tool call that settles payment before returning data.

9 tools for live social and web surfing - each one pay-per-call.

[Cascade @cascade_fyi]

Add to Claude Code:

npx x402-proxy mcp add surf https://t.co/7Ah4FzEkCJ

Supports x402 (Base, Solana) and MPP (Tempo). Proxy auto-negotiates payment.

[Cascade @cascade_fyi]

Try it - x402-proxy handles the payment flow:

npx x402-proxy --protocol mpp -X POST \
-d '{"model":"minimax/minimax-m2.7","messages":[{"role":"user","content":"Say hi"}],"stream":true}' \
https://t.co/UDmaBiWaOP

No more paying for max_tokens your agent never uses.

[Cascade @cascade_fyi]

Your agent pays for 4096 tokens of inference and gets 47 back.

That's on-chain LLM pricing today - payment commits before the model responds. Flat rate or max_tokens ceiling, you overpay either way.

Shipped per-token streaming payments on Surf Inference using @mpp sessions on @tempo. Agent pays per output token as it streams.

[Cascade @cascade_fyi]

How:
- Open a payment channel session (on-chain, once)
- Input tokens charged upfront from actual message size
- Each streamed output token = one off-chain voucher (sub-100ms)
- Channel auto-settles when idle

10+ models (Claude, Kimi, MiniMax, Qwen, Grok). OpenAI-compatible. Same endpoint supports x402 and MPP.

[Cascade @cascade_fyi]

In 1997, HTTP 402 was invented but micropayments didn't work. Ads filled the gap.

28 years later: stablecoins solved the fixed-fee problem, and agents don't click ads.

We run 30+ API endpoints on x402 + MPP. Agents pay per call. No accounts, no attention tax. The 402 era is here.

[a16z crypto @a16zcrypto]

Curated agent marketplaces look a lot like AOL. Open payment protocols look a lot like HTTP. The last time this happened, HTTP won.

What will prevail in the age of agentic commerce?

[Cascade @cascade_fyi]

x402's facilitator actually verifies the payment signature and settles on-chain before the resource server delivers data. The server doesn't respond on "submitted" - it waits for facilitator confirmation. The state gap you're describing is real for raw on-chain payments, but x402 was designed to close it.

[Cascade @cascade_fyi]

@turriztA @coinbase The key shift isn't just "pay instead of see ads." It's that agents can discover and pay for services with zero prior relationship. No signup, no API key, no approval. HTTP 402 turns every endpoint into a self-service storefront. That's what makes it HTTP-like.

[Cascade @cascade_fyi]

@AgentEconoemy Real agent traffic. We log every facilitator settlement, monitor end-to-end latency on all four endpoints. Synthetic probes miss what matters: actual payment flow under load. Current uptime 99.4% over 30 days.

[Cascade @cascade_fyi]

Which protocol do you see agents preferring for multi-step workflows? We've seen x402 work great for discrete API calls, but sessions (where payment streams over time) feel like MPP's sweet spot.

Curious if Dexter endpoints are seeing patterns in agent usage - one-shot vs long-running.

[Cascade @cascade_fyi]

@deside_app @solana @metaplex @Quantu_AI @saidinfra @fairscalexyz the agents-need-communication-rails insight is spot on.

identity (SATI/8004/SAID) + credibility (FairScale) + messaging (Deside MCP) = the stack for agents that can actually interact with humans and each other in a trust-minimized way.

this is composability done right.

[Cascade @cascade_fyi]

Shipped a fix to our Reddit service - mobile share links (/r/.../s/... redirects) were failing because we didn't handle Reddit's redirect shortcuts. User flagged it in our Telegram, fix was live within the hour.

Pay-per-use Reddit API for agents, no API keys: https://t.co/MU0pAZ4Otq

[Cascade @cascade_fyi]

@JeanClawd99 will post an update shortly

[Cascade @cascade_fyi]

@JeanClawd99 thanks for support
and yeah, we do

don't tell anybody, but Surf's mpp support is right about the corner 😎

[Cascade @cascade_fyi]

@kleffew94 thanks kevin!

[Cascade @cascade_fyi]

We built Surf - pay-per-use APIs for AI agents.

Twitter, reddit, and web search. One wallet, no API keys, no accounts.

Here's Claude Code researching x402 Week live. every API call is a real USDC micropayment on Base - you can see them streaming on the right.

https://t.co/1Eq9530keb

[Coinbase Developer Platform🛡️ @CoinbaseDev]

x402 week is live 🔥

A full week of partnerships, new integrations, feature drops… and the spotlight on YOU.

Tune in on X this Friday for a packed 402-minute (that’s 6.7 hours) event featuring builder spotlights where you can showcase your x402 build. 🧵

[Cascade @cascade_fyi]

@twitdotsh @base @solana @x402 sure

[Cascade @cascade_fyi]

Introducing Surf, pay-per-use API layer for agents on @Base and @Solana.

Social data, web crawling, and inference.
No accounts. No API keys.
Pay per call in USDC via @x402.