> For the complete documentation index, see [llms.txt](https://assay.gitbook.io/assay-docs/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://assay.gitbook.io/assay-docs/welcome/readme.md). # Welcome to Assay What Assay is, the problem it answers, and where to start reading. **Where:** the code is at [github.com/trudransh/Assay](https://github.com/trudransh/Assay). The contracts run on Monad mainnet (chain 143), with a copy on testnet (chain 10143) for free experiments. Assay gives every AI response a signed receipt. The host signs it, you can co-sign it with a passkey, and Monad checks both signatures onchain. Open verifiers grade hosts against the lab's own endpoint, and you choose which verifiers you trust. ## The problem in numbers On 26 Sep 2026 OpenRouter's public API listed the following for open-weight models. Every number comes from the [hosting report](/assay-docs/resources/hosting-report.md). | Measure | Value | | ------------------------------------------------------------------ | ------------------------------------------------------ | | Distinct endpoints for 140 open-weight models | 735, from 56 providers | | Endpoints whose precision field reads `unknown` | 289 (39%) | | Endpoints that declare a 4-bit format | 107 (15%) | | GLM-5.3 endpoints | 37, of which 16 state no precision and 8 declare 4-bit | | GLM-5.3 output price range | $1.19 to $8.80 per million tokens | | DeepSeek V4 Flash on GPQA, lab API vs. one host (measured by Olly) | 90% vs. 75% | The text you get back carries no proof of who served it. The checks that exist today stay inside the platform that ran them. ## What a receipt proves | Question | How Assay answers it | | -------------------------------------------------- | --------------------------------------------------------------------- | | Which host served this response? | The host's ES256 signature, tied to its ERC-8004 identity. | | Which model did the host claim? | The `model` field is inside the signed body. | | Was this receipt in a batch the host committed to? | A Merkle proof against a root anchored in `ReceiptAnchor`. | | Did the host pass its last audit? | `gradeOf` in `VerifierRegistry`, filtered by the verifiers you trust. | | Who asked? | An optional passkey co-signature, checked onchain. | A receipt proves who served which bytes and what they claimed. It does not prove which weights ran. Grades from verifiers cover that gap. ## Where to start | You want to | Read | | ------------------------------------------------- | ----------------------------------------------------------------------------- | | Verify a receipt on your machine in a few minutes | [Quickstart](/assay-docs/getting-started/quickstart.md) | | Learn the five ideas everything else uses | [Core concepts](/assay-docs/getting-started/core-concepts.md) | | Find a contract address or an ERC-8004 id | [Network and contracts](/assay-docs/getting-started/network-and-contracts.md) | | Call the SDK, the host or the contracts | [For developers](/assay-docs/for-developers/sdk.md) | | See what can go wrong and which test covers it | [Threat model](/assay-docs/security/threat-model.md) | ## Live today | What | Where | | --------------------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------- | | Two hosts on Monad mainnet: Gemma 4 31B (agent 10278) and Kimi K2.6 (agent 10316), every response signed and anchored | [Network and contracts](/assay-docs/getting-started/network-and-contracts.md) | | Grades from verifier 10279 against the lab's own endpoint. Both hosts are re-graded twice a week | [Verifiers and grades](/assay-docs/how-it-works/verifiers-and-grades.md) | | The Kimi K2.6 hosts on OpenRouter graded against Moonshot's endpoint (20 Oct: 15 passed 32/32, one was rate-limited) | [grades\_20261020T135810Z.json](https://github.com/trudransh/Assay/blob/dev/docs/evidence/grades_20261020T135810Z.json) | | The SDK on npm: `npm install assay-receipts` | [SDK reference](/assay-docs/for-developers/sdk.md) | | The web app: ask, verify, grades | [assay-ten-xi.vercel.app/app](https://assay-ten-xi.vercel.app/app/) | What changed since the Metropolis submission and what comes next is on the [roadmap](/assay-docs/resources/roadmap.md). Next: [Quickstart](/assay-docs/getting-started/quickstart.md) --- # Agent Instructions This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com. ## Querying This Documentation If you need additional information that is not directly available in this page, you can query the documentation by asking a question. Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters: ``` GET https://assay.gitbook.io/assay-docs/welcome/readme.md?ask=&goal= ``` `ask` is the immediate question: it should be specific, self-contained, and written in natural language. `goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case. The response will contain a direct answer to the question and relevant excerpts and sources from the documentation. Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.