Gemini Embedding by Google

# Fix list: Gemini Embedding

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/gemini-embedding, the October 2026 research run, assessed 1 October 2026. Grade BB, 71 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 Gemini Embedding: 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. Payments & pricing, 30 out of 100, up to 8.8 more on the total

Why it scored 30: No x402, MPP or L402 (0). Per-token prices are public without a login, $0.15 per million for gemini-embedding-001 on the Vertex AI pricing page, and $0.20 per million text tokens for gemini-embedding-2 as read from Vertex last week and matched by OpenRouter's listing of Google's price (20). The Gemini API has a free tier with no card, but its pricing section for the embedding models didn't load for us, so we couldn't confirm gemini-embedding-2 is on it (10 of 20). A person signs in with a Google account and creates the key (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.

## 2. Reliability, 65 out of 100, up to 7 more on the total

Why it scored 65: AI Studio has a status page for the Gemini API and Google Cloud's service health dashboard keeps product history for Vertex AI (20). The Cloud dashboard shows no Vertex AI or Gemini incident between July and September 2026, but the AI Studio page renders client-side and we couldn't read its history, so half credit between clean and unreadable (15 of 30). No published rate limits for the embedding models. The docs send you to the AI Studio dashboard and print only the batch queue limits, 500,000 enqueued tokens at tier 1 (5 of 15). The troubleshooting page gives exponential backoff with jitter, names 429 RESOURCE_EXHAUSTED and 503 UNAVAILABLE as retryable and 400, 402 and 403 as not, and the Python SDK retries transient errors four times (15). Google's Gemini SLA on Vertex AI covers only the generateContent and streamGenerateContent methods, and the Vertex AI SLA lists training and custom prediction, so nothing covers embedContent (0). gemini-embedding-2 is Stable on the model page (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.

## 3. Security & auth, 70 out of 100, up to 5.3 more on the total

Why it scored 70: Keys live in a Google Cloud project and can be restricted to the Gemini API and to IPs, referrers or apps, the docs give a rotate-then-disable routine for leaks, and no current doc puts the key in a URL. No permission scopes below the API on the Developer API route, while Vertex AI uses OAuth with IAM (25 of 30). A key restricted to the Gemini API still reaches files, caches and tuned models, so least privilege below that needs Vertex IAM roles (15 of 20). Returns vectors only (10). Opt-in request logs in AI Studio for billed projects, per the Gemini API listing's check, and Cloud Audit Logs on Vertex AI (15). security.txt is valid per the listing's provenance check, and we didn't confirm a certification that names the Developer API in this run (5 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. Agent ergonomics, 86 out of 100, up to 2.3 more on the total

Why it scored 86: output_dimensionality takes any size from 128 to 3072 and truncated vectors come back normalised, but output is float only (20 of 25). batchEmbedContents, the Batch API at half price, and documented per-request caps for text, images, audio, video and PDF pages (15 of 20). An API errors page lists the statuses, and the troubleshooting page says which to retry and which to fix, though we didn't read every message (16 of 20). Embedding calls are stateless, and the docs give backoff with jitter and a retry cap (20). Two fields needed for a call, official SDKs in Python, JavaScript, Go and Java (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.

## 5. Maintenance & community, 75 out of 100, up to 2.2 more on the total

Why it scored 75: gemini-embedding-2 went GA on 22 April 2026 per the changelog, 162 days ago (10). 18 dated changelog entries between 1 June and 22 September 2026 (20). python-genai has 190 open issues and 103 open pull requests, and every one of the newest twelve open issues carries a priority and type label, several marked awaiting user response (20 of 25). Current official SDKs, google-genai 2.25.0 on 22 September 2026 and @google/genai (15). CI on the SDK repositories, Python 3.10 to 3.14 supported (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.

## 6. Schema & documentation, 89 out of 100, up to 1.8 more on the total

Why it scored 89: A public Google API Discovery document for the Generative Language API (revision 20260930) defines EmbedContentRequest, EmbedContentConfig and batch requests (25). llms.txt with .md.txt Markdown copies of every page, including both embedding models (10). The guide says which prefix to use for queries, documents, classification and clustering, that 768, 1536 or 3072 dimensions are recommended, and warns that 001 and 2 vectors can't be mixed (16 of 20). Typed request schema, but on gemini-embedding-2 the task goes in a free-text prefix inside the content rather than an enum, and the schema still carries a taskType field the docs say can't be used with this model (11 of 15). curl examples for text, images, output size and batch, and separate API errors and troubleshooting pages (12 of 15). Dated changelog and versioned model ids (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.

## 7. Transparency & trust, 80 out of 100, up to 1.8 more on the total

Made of editorial 60, provenance 100.

Why it scored 80: Closed service under the Gemini API terms, SDKs Apache-2.0 (15). The terms say paid-tier data isn't used to improve products and free-tier data is, which the pricing and logs pages repeat, but abuse-monitoring retention on paid use has no number and zero retention is Vertex-only (20 of 30). The deprecations page lists announcement and earliest shutdown dates per model, gemini-embedding-001 until 14 May 2028, but states no minimum notice period (15 of 20). Vertex AI regions are documented, the Gemini API terms allow processing in any country where Google has facilities, and we didn't check a subprocessor list (10 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.

## 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.

- The GA date. The changelog says 22 April 2026, Google's developer blog says 30 April 2026.
- Whether gemini-embedding-2 is on the Gemini API free tier, and its Developer API price, since the pricing page section didn't load.
- What the API does if task_type is sent to gemini-embedding-2. The docs say it can't be used, not whether it's rejected or ignored.
- The AI Studio status page history for the Developer API, which we couldn't read.

## Weaknesses

- $0.20 per million text tokens, against $0.02 for OpenAI's small model
- 8,192 input tokens and float output only
- No reranker on the Gemini API
- Rate limits for embedding models are only visible in the AI Studio dashboard
- Free-tier data is used to improve Google products, and zero retention is Vertex-only

## 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.

- Don't send task_type to gemini-embedding-2. Prefix the text instead, `task: search result | query: ...` for queries and `title: ... | text: ...` for documents
- Ask for output_dimensionality 768 unless you need 3072. Google recommends 768, 1536 or 3072, and the shorter vectors come back normalised
- Use batchEmbedContents for indexing, and the Batch API for anything large, at half price
- Cap a request at 6 images, 120 seconds of video, 180 seconds of audio and one 6-page PDF. Split longer media first
- Don't mix vectors from gemini-embedding-001 and gemini-embedding-2 in one index

## What the review panel asked for

- Publish embedding limits
- Remove taskType from the schema or mark it unsupported
- Print embedding rate limits in the docs

## 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.