Crypto Bear Market Startup Guide Part 2: Token Hubs for AI Access

Original | Odaily Planet Daily (@OdailyChina)

Author | Wenser (@wenser 2010)

After exploring the pre-market spread market for crypto stocks in "The Crypto Bear Market Startup Guide, Part One," Kalshi completed a new funding round at a $22 billion valuation, significantly elevating its market position.

The second direction of this series focuses on a more everyday, higher-frequency, and often overlooked niche area—AI Token intermediaries—that even crypto-native users who regularly interact with AI tend to ignore.

This niche itself is not new. Over the past two years, numerous intermediary services across China and abroad have emerged, marketing themselves with the hook of "low-cost APIs," from 9.9-yuan monthly cards on Xianyu to widely whispered-about "reliable channels" within developer communities—this business has already grown far beyond what most people imagine. However, from the perspective of the crypto market, it has two severely underestimated dimensions: first, directly purchasing AI tokens with cryptocurrency represents an underdeveloped structural entry point; second, bundling and selling domestic models such as Qwen, Kimi, GLM, and Minimax to overseas users is an underutilized "reverse export" pathway.

In later sections, we will also mention Alex Atallah, the elite entrepreneur who decisively exited OpenSea in 2022 and subsequently founded OpenRouter, and his cross-industry transformation. Not because the project itself is particularly legendary, but because it opens up a newly emerging, severely underestimated business model for crypto entrepreneurs—the yet-to-be-fully-connected bridge between Crypto Tokens and AI Tokens.

The AI Token relay is essentially an API aggregation and forwarding layer. Users obtain a unified key through the platform, which then forwards requests on their behalf to official channels such as OpenAI and Anthropic.

The demand is real: bypassing the credit card registration barrier and reducing integration costs. It may seem like a low barrier, but the underlying complexities run deep.

According to testing by the research team on 17 third-party API platforms, 45.83% exhibited "identity mismatch"—users paid for GPT-4 pricing but actually ran inexpensive open-source models, with performance differences of up to 40%, unnoticed by most users.

This explains why many ultra-low-cost platforms frequently disappear. It’s not active fraud—it’s that their upstream account pools suffer mass suspensions, causing their cost structure to collapse instantly. The large-scale account suspension wave triggered by the Claude protocol upgrade in March 2026 was a classic example of this chain reaction.

Three types of sources: “white goods” obtained through legitimate corporate contracts, “gray goods” from bulk-registered account pools, and “black goods” from black-card top-ups or hacked accounts. The vast majority of ultra-low-price platforms rely on the latter two.

Users seek low prices, but behind low prices are unstable supply sources and hidden data risks. This contradiction currently has no solution.

OpenRouter is the most noteworthy case in this space. Founder Alex Atallah, co-founder and former CTO of OpenSea, holds a Stanford CS background and is an alum of both YC and HF0, among Forbes’ first NFT billionaires. In 2018, he co-founded OpenSea with Devin Finzer, achieving one of the most iconic wealth accumulations in crypto history by 2021. As the NFT market entered a prolonged downturn in 2022, Atallah shifted focus to AI infrastructure.

From the "unified trading layer" of NFT markets to the "unified routing layer" of LLMs, the product intuition remains consistent—building standardized aggregation entry points atop fragmented supply sides.

OpenRouter now integrates with over 60 inference providers and 300+ models, serving more than 4.2 million users worldwide and connecting to over 250,000 applications. A unified OpenAI-compatible interface enables developers to access any major model with minimal friction.

In addition, crypto payment pathways are also a rarely explored area.

The Openrouter platform has also made it possible to purchase Credits with cryptocurrency via Coinbase Business Checkouts, allowing users to directly top up with USDC/ETH on-chain without going through traditional banking channels. Of course, crypto payment channels incur an additional fee of approximately 5%, but for users seeking to avoid traditional payment friction, this premium is acceptable.

OpenRouter,APIMart.ai,cabbagewwc.com represent three distinct approaches in the current赛道.

OpenRouter follows a "crypto-native + global developers" approach, with its core strengths being compliance and founder credibility. APIMart differentiates itself through broad multimodal coverage and deep integration with Chinese models, including Qwen and ByteDance series, making it especially favorable for reverse outbound strategies. cabbagewwc represents a domestic developer hub, specializing in localized operations and RMB-denominated services, positioning it as the closest link to domestic model suppliers.

Together, these three elements form a complete value chain spanning sourcing, protocol aggregation, and crypto payments. Currently, no player has fully integrated all these stages.

If crypto payments represent "entry differentiation," then reverse outbound expansion belongs to "supply differentiation."

Following the trend that refined processing yields significantly higher profits than coarse processing, the profit margin for the latter is naturally even more astonishing.

Based on data from early 2026: Qwen3.5's price for one million tokens is as low as 0.8 RMB, approximately $0.11, which is 1/18th the cost of Gemini 3 Pro and more than 27 times lower than Claude Sonnet 4.6's $3 input price.

GLM-5 achieved a score of 77.8% on the SWE-Bench Verified programming benchmark, surpassing Gemini 3 Pro and approaching Claude Opus 4.5, while its API price is only a fraction of the latter. Kimi K2.5 has generated more revenue in nearly 20 days than its entire projected 2025 annual total.

These models have very limited availability overseas: registration barriers, payment restrictions, language interfaces, and information gaps among overseas developers regarding the capabilities of Chinese models form an invisible barrier to entry.

The opportunity for reverse outbound transit hubs lies right here.

For specific implementation, consider bulk-purchasing model API quotas in RMB within China, exposing an OpenAI-compatible interface through a protocol translation layer, and selling it to overseas developers and startups priced in USDT/USDC. Alibaba Cloud’s Bailian Coding Plan provides a cost reference: bundling four major models—Qwen3.5, GLM-5, MiniMax M2.5, and Kimi K2.5—new users can obtain 18,000 request credits for just 7.9 RMB in their first month. Pricing this offering in USD for international markets offers substantial profit potential.

Don't blindly cheer for bull runs. Before this business truly takes off, several barriers must be acknowledged.

Capital requirements. Bulk purchasing domestic model API quotas, building a technical forwarding layer, and maintaining overseas servers and crypto payment channels all require upfront capital investment. More critically, liquidity management—exchange rate fluctuations and friction in deposit and withdrawal between crypto receipts and RMB payments—can easily lead to cash flow issues without a mature funding rotation solution.

Resource channels. A stable domestic model API procurement channel is a core asset. Official channels mean establishing business partnerships with model vendors or cloud platforms, which requires time and qualifications; the account pool approach faces ongoing risks of account bans and compliance issues. At the same time, the ability to reach overseas users is equally essential—cold starting on channels such as Twitter/X, Reddit, Discord, and Telegram presents a real barrier for teams without experience in overseas community operations.

Legal compliance. Risks arise from both ends. The model’s terms of service restrict resale; the vast majority of major vendors explicitly prohibit commercial resale of APIs, leaving the account pool model in ongoing legal breach risk. Data security and cross-border compliance: selling domestic model services to overseas users involves compliance requirements for data outbound transfers, which require careful evaluation under current regulatory conditions. Carno’s payment processing may also trigger VASP licensing requirements in certain jurisdictions.

Another point: There is ample evidence within the industry that some intermediary platforms bundle and sell user prompt data for model training. This is not only a legal risk, but also a business time bomb that, if exposed, would immediately destroy user trust.

The barrier is not in technology, but in resource integration and risk management. Teams that can simultaneously master these four areas—domestic low-cost model procurement channels, OpenAI-compatible protocol conversion, crypto payment gateways, and overseas user operations—are virtually nonexistent in today’s market.

This is both an opportunity and a practical challenge.

From Alex Atallah pivoting to found OpenRouter after the NFT downturn, to a group of domestic developers quietly building relay stations serving tens of thousands of users, the very existence of this business is a real-world answer to the question: “What can survive in a bear market?”—it doesn’t rely on token speculation or fundraising narratives, but generates real revenue through actual API call volumes.

Deep water doesn't mean you can't swim. The key is to measure how deep the water is before getting in.

Recommended Reading

Crypto Bear Market Startup Guide, Part 1: The Pre-Market Spread Market for Crypto Stocks

From 5 jiao per kilowatt-hour of Chinese electricity to 45 yuan API export packages: Token is becoming a new unit of currency

After going through various sketchy intermediaries, I slammed my table! Here's a guide to building your own intermediary from scratch!