Nomic Embed

# Fix list: Nomic Embed

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/nomic-embed, the October 2026 research run, assessed 8 October 2026. Grade D, 49.2 out of 100.

This is everything the published grade says the listing lacks, the biggest possible gain to the total first. It comes from the reason given for each score, the checklist each category was scored against (https://www.anchorterminal.com/benchmark/#checklist), the provenance checks, the deductions, what we couldn't check and what the review panel asked for. A fix counts at the next check, once it's public.

For a coding agent working on Nomic Embed: work through the items below in the product, its docs and its public pages. Each category gives the reason for its score, with the points each checklist item earned, and the checklist itself, so the gap is the items that earned less than their points. Change the product, not the wording, and keep a note of what you changed and where it's published.

## 1. Reliability, 38 out of 100, up to 12.4 more on the total

Why it scored 38: Graded as a hosted service, on the embedding endpoints at api-atlas.nomic.ai. status.nomic.ai exists but its components are the Nomic Platform in four regions, with none for api-atlas.nomic.ai (5 of 20). Its history from July to October 2026 lists six incidents on the platform, none naming embeddings, so there is no readable record for this endpoint (5 of 30). The limit is published as 1,200 requests per five-minute rolling window per IP address (15). Retry-After on a 429 is documented for the Platform v0 API only. For embeddings the evidence is the Python client, which retries 429 and 5xx up to five times with exponential backoff, and the calls only read (8 of 15). No SLA found, and the Individual terms promise reasonable commercial efforts (0). The embedding pages sit in the main API reference and the endpoint answers, but the API is versioned v0.57.0, the legacy reference warns its endpoints are unstable, and the current documentation index omits Atlas (5 of 10).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-reliability):

Hosted APIs, MCP servers, models and platforms.

- 20, a public status page with component history (Statuspage, Instatus, BetterStack or the vendor's own).
- 0 to 30, the incident record for the last 90 days on that page. 30 for a clean record or trivial incidents only, 20 for minor incidents only, 10 for one major outage (an hour or more of a core API down, or errors across the board), 0 for several. 5 when there's no history we could read, and the note says so.
- 15, rate limits documented with numbers.
- 15, documented 429 or overload handling (Retry-After, backoff guidance), and idempotency keys or safe-retry guidance where writes are involved.
- 10, an SLA published for any paid tier.
- 10, the surface agents use is generally available, not beta or preview.

Local packages, SDKs, frameworks and stdio MCP servers.

- 20, installs from an official package with supported runtimes stated.
- 25, a public CI and test suite, passing on the default branch.
- 0 to 25, open crash or regression issues relative to activity (25 for few and handled, 0 for many, old and unanswered).
- 15, semver discipline and breaking changes called out in a changelog.
- 15, version 1.0 or later, or declared stable.

Protocols are read from their reference implementations, the public facilitators or servers, spec stability and test vectors.

## 2. Payments & pricing, 20 out of 100, up to 10 more on the total

Why it scored 20: No x402, MPP or L402 (0). www.nomic.ai/pricing lists only the Nomic Platform plans. The Atlas web app's script lists $1 per 10M tokens and $1 per 50,000 images, but atlas.nomic.ai/pricing renders only in a browser and we couldn't read it as a page (10 of 20). The same script lists a starter plan with 10M tokens free. Whether a card is needed, and whether new accounts are accepted, wasn't confirmed (10 of 20). Sign-up is in a browser, and the key-creation endpoint needs an existing login (0).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-payments):

The published rubric, also on the [x402 page](https://www.anchorterminal.com/x402/).

- 40, a machine payment protocol (x402, MPP or L402) on the tool's own endpoints. 10 to 30 when it covers only some endpoints or only goes through a third party, and the note says which.
- 20, per-call or per-unit pricing published without a login. 10 for public plan-only pricing, 0 for "contact sales" or prices behind a login.
- 20, a free tier or trial that doesn't need a card.
- 20, autonomous onboarding, meaning an agent can get access without a person signing up in a browser (keyless use, x402, a programmatic key API).

Payment platforms and agent wallets rarely charge for their own API over a machine protocol, so the first line has steps for them, and the highest one that applies counts. 40 when x402, MPP or L402 runs on all their own endpoints, 30 when it runs on part of their own API, 25 when their merchants can accept one, 20 for running a facilitator, 15 for paying as a buyer, and 0 when the only protocol is their own. Merchant acceptance sits above a facilitator because the platform's own customers can charge agents through it, while a facilitator settles for sellers who wire up the protocol themselves. The counter-argument (a facilitator does more for the protocol as a whole) has a point. Each note says which step applied.

Open-source software you run yourself is scored on its hosted or paid option if it has one. A free, self-hosted package with nothing to buy gets 20, 20 and 20 for the last three lines, and 0 to 40 for the first only if it ships a payment protocol.

## 3. Security & auth, 62 out of 100, up to 6.7 more on the total

Why it scored 62: Bearer API keys created and deleted in the Atlas dashboard or through key endpoints, scoped to an organisation, a dataset or a user. No rotation or expiry documented (25 of 30). The embedding calls only compute vectors, and Atlas roles from Owner to Guest limit who can create keys, but a key scoped to an organisation also reaches dataset endpoints (12 of 20). The endpoint returns vectors and token counts, no untrusted content (10 of 15). The API has organisation usage endpoints. No per-call log for Atlas was found, and the audit log Nomic documents belongs to the Platform v0 API (5 of 15). The security page gives security@nomic.ai with acknowledgement in five business days, and says Nomic is SOC 2 Type II certified with annual penetration tests, the report under NDA. That page is written for Business and Enterprise platform accounts. No security.txt and no bug bounty found (10 of 20).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-security):

- 0 to 30, the credential model. 30 for OAuth 2.1 with scopes, or scoped and revocable keys with rotation. 20 for plain revocable API keys. 10 for one all-powerful key. 10 off when a secret can travel in a URL query string as a documented option.
- 0 to 20, read-only or least-privilege modes, and confirmation or approval for destructive actions.
- 0 to 15, prompt-injection posture where the tool returns untrusted content (documented mitigations or guidance). A tool that returns no untrusted content gets 10.
- 0 to 15, audit logs or per-call visibility for the operator.
- 0 to 20, a security programme. security.txt or a disclosure policy, a bug bounty, SOC 2 or ISO 27001, advisories handled in public.

Models are read for retention, whether API data trains models (and whether that's off by default), zero-retention options and certifications. Frameworks for telemetry defaults, approval hooks, guardrails and sandboxing.

## 4. Maintenance & community, 28 out of 100, up to 6.3 more on the total

Why it scored 28: The Atlas API reports v0.57.0 with no date. /v1/version dates its newest component, convert, 19 June 2026, 111 days ago, and the embedding model cards for v1 and v1.5 were last changed on 7 April 2026 (10 of 30). No dated API changes or client releases in the last 90 days (0 of 20). The Python client's repository has 50 open issues, with the newest issue activity on 19 March 2026, and a community Slack is linked from the docs (5 of 25). Official clients exist for Python (nomic 3.9.0, 11 November 2025) and TypeScript (@nomic-ai/atlas 0.12.0, last pre-release 16 September 2025), neither released in 2026 (8 of 15). The Python repository has a CircleCI configuration, a changelog and typed packaging, with its last commit on 11 November 2025 (5 of 10).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-maintenance):

- 0 to 30, time since the last release, or the last published model or API change for a closed service. 30 within 30 days, 20 within 90, 10 within 180, 0 older.
- 20, at least three releases or dated changelog entries in the last 90 days.
- 0 to 25, responsiveness. Issues and pull requests answered on GitHub (the open issues and how recent the replies are). For closed services, a public changelog and a support or community channel that answers, 0 to 15.
- 15, presence in the official MCP registry under a verified namespace (MCP servers), or current official SDKs (APIs and models).
- 10, package health, current dependencies and CI.

Models are read for deprecation notice periods and model churn rather than release counts.

## 5. Schema & documentation, 65 out of 100, up to 5.7 more on the total

Why it scored 65: OpenAPI 3.1 document for the Atlas API, v0.57.0, public at api-atlas.nomic.ai/v1/api-reference/openapi.json (25). docs.nomic.ai/llms.txt exists but lists no embedding page, and the Markdown copies it describes return 404 for the Atlas and reference pages (3 of 10). The reference and the generate-embeddings guide say which task_type fits queries, documents, classification and clustering (14 of 20). model and long_text_mode are enums and texts is required, but task_type is a free string in the spec and dimensionality has no bounds (10 of 15). curl, Python and requests examples, with only a 422 validation error documented. We saw an undocumented 403 with the body Unauthorized user (7 of 15). The spec carries a version and /v1/version answers, but there's no API changelog and the docs page and the spec disagree on the model list (6 of 15).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-schema):

APIs and MCP servers.

- 25, a machine-readable contract (a public OpenAPI file or similar; for MCP, typed JSON Schema inputs on every tool).
- 10, llms.txt or Markdown docs served for agents.
- 0 to 20, descriptions that say what a tool is for, when to use it and when not to, read from the tool definitions in the source or the API reference.
- 0 to 15, typed inputs with enums, constraints and required fields, and no free-form JSON blobs.
- 0 to 15, examples and documented error responses.
- 15, versioning and a public changelog.

Models are read from the API reference, the OpenAPI file, llms.txt, the structured-output and tool-use docs and the model cards. Frameworks from docs a model can follow, typed interfaces, examples and the API reference.

## 6. Agent ergonomics, 69 out of 100, up to 5 more on the total

Why it scored 69: dimensionality from 64 to 768 on nomic-embed-text-v1.5, floats only, with no binary or base64 output (18 of 25). Batched texts, long_text_mode and max_tokens_per_text control input size, though the Python client caps a request at 10 texts since 3.5.3 (14 of 20). One documented error, a 422 with field locations, and a bare 403 for a missing key (8 of 20). Stateless read-only calls that are safe to repeat, with backoff built into the Python client and no idempotency key needed (16 of 20). Only texts is required, and there are official Python and TypeScript clients, both without a release in 2026 (13 of 15).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-ergonomics):

- 0 to 25, context cost. For MCP, the number and size of the tool definitions (25 for ten or fewer compact tools, 15 for 11 to 30, 5 for more than 30, plus up to 10 back for toolsets, dynamic loading or read-only subsets). For APIs, whether responses can be sized (field selection, limits, summaries).
- 20, pagination, filtering and output-size controls.
- 20, actionable, documented error responses, codes and messages an agent can recover from.
- 20, idempotency or safe retries, and for MCP the `readOnlyHint` and `destructiveHint` annotations.
- 15, sensible defaults, few required parameters, and official SDKs in at least two languages.

Models are read for tool use, structured output, prompt caching, context length, batch and SDKs. Frameworks for how much code and how many defaults a tool-calling agent with MCP needs.

## 7. Transparency & trust, 46 out of 100, up to 4.7 more on the total

Made of editorial 47, provenance 45.

Why it scored 46: The hosted API is closed, while the weights for the text, code and vision models are Apache-2.0 on Hugging Face and the Python client declares Apache, though its repository has no licence file (25 of 30). No published terms or privacy policy name Atlas or the embedding API. The privacy policy covers www.nomic.ai and app.nomic.ai, the Business and Enterprise terms forbid training on Customer Data, and the Individual terms allow use to improve the technology behind it, so the position for embedding inputs is unclear (8 of 30). The Individual terms promise 30 days' notice before endpoints change and six months of support for the previous API version, where an order includes the API. No dated notices for Atlas found (8 of 20). The security page lists hosting regions and subprocessors for the platform, and the privacy policy says servers are in the US. Nothing specific to api-atlas.nomic.ai (6 of 20).

The checklist (https://www.anchorterminal.com/benchmark/#checklist-transparency):

- 0 to 30, source availability and licence clarity. 30 for open source under an OSI licence, 15 for closed with clear terms, 0 for unclear terms.
- 0 to 30, data handling and retention statements that agree with each other (privacy policy, DPA, retention periods, subprocessors).
- 0 to 20, a deprecation policy or notices with dates.
- 0 to 20, telemetry disclosed with an opt-out (local software), or subprocessors and data locations disclosed (hosted).

The other half of Transparency and trust is the provenance score, computed from checked facts (below). The category score is the mean of the two.

Provenance checks not met in full (half of this category, computed from checked facts):

- Domain age: nomic.ai, no registry record we could read (0 of 15)
- Terms of service: not found (0 of 10)
- Privacy policy: not found (0 of 10)
- Status page: not found (0 of 10)
- security.txt: not found (0 of 10)

## What we couldn't check

What we couldn't read counted as absent. Publishing it on a page a plain HTTP fetch can read (not only in a browser) lets the next check count it.

- unchecked: whether atlas.nomic.ai still accepts new sign-ups and issues API keys to them. The app renders only in a browser. Its script contains Log In and Sign Up buttons
- unchecked: the rendered Atlas pricing page. The app's script lists a starter plan with 10M tokens free and $1 per 10M tokens on paid plans, which is where the listed price comes from. Whether a card is needed wasn't established
- Which terms and privacy policy govern the Atlas embedding API. None of Nomic's three published legal documents names it
- Whether Nomic trains on text sent to the embedding endpoint. The Business and Enterprise terms forbid training on Customer Data, while the Individual terms allow use to improve the technology behind it
- The docs page lists nomic-embed-text-v1 and v1.5 as the model values, while the OpenAPI document also lists nomic-embed-text-v2, nomic-embed-code, gte-multilingual-base and all-MiniLM-L6-v2. Which of these the endpoint serves wasn't tested, since we had no key
- unchecked: the 429 response of the embedding endpoint. Retry-After is documented for the Platform v0 API only
- Whether Nomic plans to retire the Atlas API. No dated notice was found, but the legacy reference warns its endpoints may be removed and the current documentation index omits Atlas
- The lead said the API page names only v1 and v1.5, which is right for the docs page. The OpenAPI document names six text models and two vision models

## Weaknesses

- docs.nomic.ai/llms.txt and www.nomic.ai now describe a product for architecture, engineering and construction firms, and the documentation index no longer lists the embedding pages
- No rendered public page states a price for the endpoint. The $1 per 10M tokens figure comes from the Atlas web app's script
- status.nomic.ai has no component for api-atlas.nomic.ai, and no SLA was found
- The published terms and privacy policy cover the Nomic Platform at app.nomic.ai and don't name Atlas or the embedding API
- The Python client's last release was 3.9.0 on 11 November 2025, and the API reference documents only a 422 error

## What costs an agent a turn today

The notes we give agents before they call it. Each one is a workaround an agent shouldn't need.

- Set task_type to search_query for queries and search_document for stored text. The default is search_document
- Name the model in every request. The API defaults to nomic-embed-text-v1, while the Python client defaults to nomic-embed-text-v1.5
- Keep under 1,200 requests per five minutes per IP address. The Python client sends at most 10 texts a request
- Set long_text_mode to truncate or mean. Texts over 8,192 tokens are averaged across chunks by default on the API
- Pass dimensionality only with nomic-embed-text-v1.5, between 64 and 768

## When it's done

Send what changed and where it's published as a dispute (https://www.anchorterminal.com/builders/#disputes, or `POST https://www.anchorterminal.com/api/v1/contact` with `"kind": "dispute"`). Disputes are answered in public, and the listing is checked again by the same checklist. Paying for an audit or a listing claim changes nothing here.