Orbio.so is an AI inference relay platform that enables developers to access 447 LLMs—including Claude Fable 5.1, GPT-6 Astra, and Gemini 3.8 Flash—using a single OpenRouter-compatible API key. Its official website (orbio.so) states it adds under 50 ms latency per request, stores no prompts, and charges per token at discounted rates (e.g., 80% off list price for several models). Users deposit USD to fund usage; recent deposits shown on the live dashboard range from $5 to $500, with discount estimates varying between ~60% and ~79% of list-value equivalent. The ORBIO token is contract-deployed on the Robinhood Chain (EIP-155:4663) at address 0xaa07a0e9209e16ac99708c3ec70159c6ef3128a3, and its native explorer is Robinhood Chain Blockscout. According to two independent API sources, ORBIO’s stated utility includes distributing half of trading fees as inference credits to token holders—but neither source confirms active fee generation, credit distribution, or on-chain transaction volume. Key evaluation questions remain: Is there verifiable on-chain economic activity (e.g., transfers, staking, fee accrual) tied to the ORBIO contract? Does the relay’s claimed usage discount reflect real-time arbitrage or subsidized capacity—and is that capacity independently measurable? How is the circulating supply reconciled, given one source reports 0 in circulation while another reports 950M?
Orbio.so positions itself as a cost-optimized AI inference relay, offering up to 80% discounts on model calls by aggregating unused capacity from OpenRouter keys and $ORBIO token holders. Its website (web_1) documents the technical flow—unchanged prompts, provider-authenticated responses, no prompt storage—and shows live deposit activity and usage analytics. However, the core thesis—that discounted credits are sourced from token holders and key suppliers—is asserted but not verified: no mechanism for holder participation, no on-chain proof of credit minting or distribution, and no public documentation of supplier onboarding or vetting appears in the evidence. The tokenomics claim—“Half of trading fees are converted into inference credits and paid to holders, in dollars”—(api_2) lacks supporting detail on fee collection, conversion logic, or payout timing. Critically, the circulating supply is reported as 0 (api_1), while total supply is stated as 950M tokens—yet no evidence confirms whether or not tokens are deployed, transferable, or integrated with the relay. Governance rights, decision control, and loss allocation (e.g., if a supplier’s key fails or misroutes) remain entirely undocumented. No audits, team disclosures, or operational transparency (e.g., uptime logs, incident reports) are present. The project functions as a frontend with plausible economics—but its incentive architecture and token-backed value proposition rest on unverified claims.
Incentive-map:
The absence of verifiable primary-thesis evidence—especially around token utility, credit sourcing, and loss-bearing parties—means the model cannot confirm alignment or sustainability. This is a borderline case: functional enough to test, but lacking the foundational transparency needed for higher confidence.
Overall score: 5/10 Confidence: Low