Microsoft-Decision-1

# Fix list: Microsoft-Decision-1

From Anchor Terminal's listing at https://www.anchorterminal.com/tools/microsoft-decision-1, the October 2026 research run, assessed 10 October 2026. Grade E, 39 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 Microsoft-Decision-1: 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, 30 out of 100, up to 14 more on the total

Why it scored 30: Azure status history has a Foundry Models component with dated incidents, earning 20 of 20 for a public status page with component history. This is a shared service component, not a page for this model. The 29 September 2026 incident in Sweden Central, from 10:03 to 15:58 UTC, with intermittent failures and HTTP 5xx, is one regional outage of about six hours, earning 10 of 30 for one major. The 30 September multi-region connectivity incident does not name Foundry Models, so it is not counted. No rate limit, 429 guidance or SLA for this model was found (0, 0 and 0). The model is in public preview, so there is no general availability award (0). Total 30.

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: The price is published without a login at $0.042 per million input tokens, with output free, earning 20 of 20. No x402, MPP or L402 was found in the announcement, the Foundry sample or OpenRouter's model page, earning 0 of 40. No free tier or trial without a card was found (0 of 20). The Foundry route needs an Azure subscription and deployment set up by a person, and OpenRouter's signup was not checked (0 of 20). Total 20.

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. Schema & documentation, 43 out of 100, up to 9.3 more on the total

Why it scored 43: No machine-readable contract for the model was found (0 of 25). OpenRouter's llms.txt indexes its Decisions SDK pages as Markdown, earning 10 of 10 for the OpenRouter route. Microsoft's post states what the model is for and what it is not for, earning 15 of 20 without a reference page. The question fields (type, instructions and criteria) are typed in Microsoft's sample, earning 10 of 15, with no published schema read. The sample shows a request and handles an HTTP error, but no documented error responses were found, earning 8 of 15. No changelog was found, and OpenRouter says weights are updated continually while the API shape stays the same, so 0 of 15. Total 43.

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, 52 out of 100, up to 8.4 more on the total

Why it scored 52: The Foundry sample uses a Microsoft Entra bearer token for a fixed resource scope, earning 20 of 30. Role assignments and token lifetimes were not read, and OpenRouter uses an API key. The model only scores inputs, with no write or destructive operation, earning 10 of 20 for least privilege, since no scoped inference permission was found. Microsoft reports tests of 5,250 requests across 11 benchmarks, including prompt injection, but the post gives no integrator guidance, earning 5 of 15. OpenRouter's Activity page shows token volume and requests, and Foundry diagnostic logs were not read, earning 5 of 15. Microsoft's security.txt names a disclosure policy and a bounty, but it expired on 23 September 2026, earning 12 of 20. Total 52.

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. Agent ergonomics, 50 out of 100, up to 8.1 more on the total

Why it scored 50: The response is one probability per option, and the caller sets the number of questions and options. This earns 20 of 25 for context cost. No pagination, filtering or per-request limits were found for this model, earning 10 of 20. No documented error codes and recovery actions were found, earning 5 of 20. No idempotency or retry guidance was found, earning 5 of 20. Three required fields, a Python sample and OpenRouter's alpha SDK in three languages earn 10 of 15, since the SDK method is alpha. Total 50.

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.

## 6. Transparency & trust, 32 out of 100, up to 6 more on the total

Made of editorial 22, provenance 41.

Why it scored 32: No licence for the hosted model is stated in the announcement or on OpenRouter's page, and no Microsoft weights repository was found, so 0 of 30 for licence clarity. Microsoft's Foundry data page states no storage in the model, no training on prompts and processing within the geography or data zone, but abuse monitoring stores prompts for human review. The page does not say whether it covers this model, so 12 of 30. No dated deprecation policy was found, 0 of 20. The US and EU Datazone deployment types disclose where processing may happen, earning 10 of 20. No subprocessor list was read. Total 22.

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):

- Legal entity named: not found (0 of 20)
- Domain age: microsoft.com, no registry record we could read (0 of 15)
- Changelog: not found (0 of 10)
- security.txt: published but past its Expires date (5 of 10)

## 7. Maintenance & community, 40 out of 100, up to 5.3 more on the total

Why it scored 40: The model was released on 9 October 2026, within 30 days, earning 30 of 30. The record holds one dated release, so there are no three releases or changelog entries in 90 days (0 of 20). For a closed service, responsiveness needs a changelog and a support channel that answers, and neither was sampled (0 of 15). OpenRouter's SDK lists the Decisions method in Python, TypeScript and Go as alpha, earning 10 of 15 for current official SDKs with the method. No CI or dependency record for the model was read (0 of 10). Total 40.

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.

- unchecked: the Foundry catalogue page at ai.azure.com is rendered by script and returned no model content. The deployment name, regions, terms and context length on the Foundry route were not read.
- unchecked: the Foundry quickstart, which Microsoft's sample names as the source of the deployment identifier and route.
- unchecked: Microsoft's product terms and any licence for Microsoft-Decision-1. None is stated in the pages read.
- unchecked: whether the Foundry data page for Models sold by Azure covers this model. The page does not name it.
- unchecked: rate limits and 429 handling for the model on either route.
- unchecked: OpenRouter's privacy and data terms, its signup and key steps, and whether it has a free allowance.
- No authenticated call was made, so the latency, request format and error claims are not verified.
- The two Microsoft posts disagree on the Xbox baseline. The Tech Community post says GPT-5 and 80 to 100 times faster. The Command Line post says GPT-6 Sol, over 14 times faster and 200 times less expensive. Neither gives a quality figure.
- The calibration statement is a stated goal, not a measured result. The 36-benchmark accuracy claim has no published scores.

## Weaknesses

- Public preview only. Microsoft's post gives no general availability date and states no licence for the hosted model
- No model-specific rate limit, retention statement or deprecation policy was found. Microsoft's Foundry data page covers Models sold by Azure in general
- The Foundry route needs an Azure subscription, a deployment of the model and a Microsoft Entra token, all set up by a person
- Microsoft describes its Xbox Research result as competitive with GPT-5 or GPT-6 Sol on quality, not better, and the two Microsoft posts name different baselines for it
- Azure status history lists a Foundry Models incident on 29 September 2026 in Sweden Central, with intermittent failures and HTTP 5xx for about six hours

## 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.

- Use OpenRouter's Decisions method, not an OpenAI chat-completions SDK. OpenRouter says chat completions SDKs will not work with this model
- On Foundry, send the request to the deployment's /providers/microsoft/v1/systemone path with a Microsoft Entra token for https://cognitiveservices.azure.com/.default, not an API key
- Take the deployment name from the Foundry quickstart before the first call. Microsoft says to confirm the route and authentication header, and the pages reviewed do not give the name
- Keep each request within OpenRouter's 32,768-token context and send only fixed options, since the model is not intended for open-ended generation
- Measure latency and calibration on your own labelled cases before relying on Microsoft's latency and 'nine times out of 10' statements

## 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.