Claude API by Anthropic

# Fix list: Claude API

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/anthropic-api, the October 2026 research run, assessed 1 October 2026. Grade BB, 77.6 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 Claude API: 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 machine payment protocol (0). Per-token prices published without a login (20). The pricing page says new users get a small amount of free credit but doesn't say whether a card is needed, so half (10, a judgement call). Keys are created in the browser Console only, and the Admin API can't create one (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, 60 out of 100, up to 8 more on the total

Why it scored 60: Status page at status.claude.com with component history (20). Its feed shows three API incidents of 80 minutes or more with elevated errors across several models in the last 90 days, 24 August (about 2.5 hours), 3 September (about 3 hours) and 22 September (about 80 minutes), plus a 59-minute all-product error spike on 29 September, so several majors (0). Rate limits published per model, 1,000 requests a minute at the Start tier (15). 429s carry `retry-after` and the SDKs retry with backoff (15). No SLA on sale. Priority Tier's 99.5% uptime target is closed to new buyers, and a target isn't a commitment (0, a judgement call). Messages API is GA (10). Graded by agents running on Claude, by the same checklist as every other listing.

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. Schema & documentation, 90 out of 100, up to 1.6 more on the total

Why it scored 90: No OpenAPI document in the docs. A Stainless-generated spec is reachable from the Python SDK's .stats.yml, which we count as half a contract (15 of 25, a judgement call applied the same way to every listing with this setup). llms.txt indexes 741 pages with Markdown twins (10). The reference documents each parameter and the models page says which model to use when (20). Structured outputs and strict tool use are GA with grammar-constrained sampling, though `minimum` and `maxLength` aren't enforced (15). Errors page lists 11 error types with a `request_id`, and examples run through the docs (15). `anthropic-version` header and dated release notes (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.

## 4. Security & auth, 92 out of 100, up to 1.4 more on the total

Why it scored 92: Model reading. Personal and service-account keys scoped to a workspace, with expiry set at creation, disable and delete, plus Workload Identity Federation for short-lived tokens. No permission scopes below the workspace, so 25 of 30 (a judgement call, OpenAI's endpoint-scoped keys get 30). No training on API data without express permission (20). Retention capped at 30 days in the privacy centre, ZDR by sales agreement, but Fable 5.1 and Mythos 5.1 require 30-day retention and are outside ZDR (12 of 15). Usage and Cost API plus a Compliance API activity feed (15). security.txt valid to 2026-12-31, HackerOne programme, SOC 2 Type II, ISO 27001 and ISO 42001 (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.

## 5. Transparency & trust, 88 out of 100, up to 1.1 more on the total

Made of editorial 75, provenance 100.

Why it scored 88: Closed service with clear commercial terms, SDKs MIT (15). Retention is stated two ways, 'not retained by default' in the docs and 'deleted within 30 days' in the privacy centre, with 30-day retention required on Fable and Mythos models, so an operator can't tell which applies to a given call (20 of 30). Deprecation page with announcement and retirement dates (20). Subprocessor list at trust.anthropic.com, processing countries named and storage stated as US (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.

## 6. Agent ergonomics, 95 out of 100, up to 0.8 more on the total

Why it scored 95: Model reading of the checklist (tool use, structured output, caching, context, batch, SDKs, errors). Tool use with parallel calls, 15 of 20 because `tool_choice` any or tool returns 400 on Opus 5.5, Sonnet 5.5 and Fable 5.1, so a tool can't be forced on the current models. Structured outputs GA (15). Prompt caching with reads at 0.1x input, 0.05x on Opus 5.5 (15). 1M context on the default model (15). Batch at half price (10). Official SDKs in Python, TypeScript, Java, Go, Ruby, C# and PHP (10). Documented error types, `retry-after` and SDK retries with backoff (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. Maintenance & community, 91 out of 100, up to 0.8 more on the total

Why it scored 91: Model reading. Sonnet 5.5 launched on 28 September and Sonnet 4.5's deprecation was posted on 30 September (30). Published minimum notice of 60 days, emailed to active users (8 of 12). One model shutdown date in the last 90 days, Opus 4.1 on 5 August (8 of 8). Dated release notes several times a week and a support centre (15 of 15). The Python SDK has 52 open issues, and a sampled strict-tool bug opened on 10 September had no reply three weeks later (5 of 10). Current official SDKs, Python 1.11.0 on 30 September (15). Supported runtimes stated, Python 3.10 to 3.14 (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.

## 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.

- Whether the free credit for new users needs a card or phone verification. The pricing page doesn't say, and the listing's `card-required` tag may be wrong
- Which retention statement governs a standard Messages call, 'not retained by default' or 'deleted within 30 days'
- Incident severity was read from the status feed's titles and timestamps. The feed doesn't label impact, so the major and minor split is our reading
- Disclosure. These grades were written by agents running on Anthropic's Claude. The judgement calls are the half credit for the SDK-hosted OpenAPI spec, 25 of 30 for workspace-scoped keys, 10 of 20 for free credit with an unstated card rule, and no SLA credit for a Priority Tier target that's no longer sold

## Weaknesses

- Three incidents of 80 minutes or more with elevated errors across several models between 24 August and 22 September 2026
- Keys are created only in the browser Console, and there's no machine payment
- `tool_choice` any or tool returns 400 on Opus 5.5, Sonnet 5.5 and Fable 5.1, and thinking can't be switched off on them
- Retention reads 'not retained by default' in the docs and 'deleted within 30 days' in the privacy centre
- No SLA on sale. Priority Tier's 99.5% target is closed to new buyers

## 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.

- Default to `claude-opus-5-5` and keep `claude-fable-5-1` for tasks that fail on Opus, at 2.5 times the price
- Don't send `tool_choice` any or tool to the 5.x models. Use auto with `strict: true` on the tool
- Put `cache_control` on the system prompt and tool list. Reads cost 0.05x input on Opus 5.5
- Wait out a 429 by its `retry-after` seconds, but a spend-cap 429 has no header and won't clear by waiting
- Move off `claude-sonnet-4-5-20250929` before 2026-11-30. Retired ids fail, they don't redirect

## 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.