Head to head · Spend management · October 2026 research run

Brex vs Mercury API

Mercury API scores 66.1 (B) on agent readiness against Brex's 64.5 (B), and leads in 3 of 7 scored categories. Brex leads on reliability and schema & documentation. Both do spend management.

Best spend management and procurement platforms for AI agents · All 36 spend comparisons

Which one, for what

Brex B

Best for A finance team already on Brex that wants an agent to read expenses and transactions, issue and lock cards, set spend limits, upload receipts and pay vendors.

Ahead on

  • Reliability, 75 against 57
  • Schema & documentation, 80 against 72

Watch for

The Expenses API update endpoint accepts only memo, so an outside agent can't set a category or custom field on an expense through it

Mercury API B

Best for A company that banks with Mercury and wants an agent to read balances and transactions, set notes and categories, upload receipts, issue virtual cards with spend limits and queue payments for human approval.

Ahead on

  • Security & auth, 82 against 72
  • Payments & pricing, 45 against 38
  • Maintenance & community, 74 against 56

Also in its favour

  • No incidents deducted, where Brex loses 3 points for them

Watch for

No rate limits with numbers and no 429 handling were found in the reviewed documentation

Why the edge

Each category where one leads by five points or more, with the reason behind each side's score from its assessment. The numbers are in the table below; this is what they stand for.

Reliability Brex +18

Brex 75

Read with the hosted lines and scored on the REST Developer API, the surface an outside agent calls with a token. status.brex.com on Atlassian Statuspage, with components and incident history (20). Its history.atom, read on 3 November 2026, lists eight entries from 5 August to 3 November. One touches the API, "Unexpected invalidation of some Developer API Tokens" on 21 September, which names some tokens, so it reads as partial. The others touch the dashboard (11 August, 30 September), the Mastercard network (15 August), the trust portal (1 and 2 September), accounting pages (15 September), notifications (2 October) and SMS to France (28 to 30 October). The API errors of 4 August (16:01 to 20:34 UTC) fall before the window. Minor incidents only (20). The 8 October reading came from the status host's /api/ path, which its robots.txt disallows, and is replaced. Limits with numbers, 1,000 requests in 60 seconds per client and account, plus daily caps on transfers, international wires and cards (15). The rate-limit guide recommends exponential backoff with randomness on 429, and every POST and PUT accepts an Idempotency-Key, required on transfers and cards. A Retry-After header isn't documented (13 of 15). No SLA found in the Platform Agreement or the Access Agreement (0). Team, Budgets, Payments, Transactions, Travel and Onboarding are at version 1.0, while Fields is beta, Accounting alpha, Expenses and Webhooks are numbered 0.1 and the MCP server is beta (7 of 10). 20 + 20 + 15 + 13 + 0 + 7 = 75.

Mercury API 57

Read with the hosted lines and scored on the REST API, the surface an outside agent calls with a token. status.mercury.com runs on incident.io with 11 components, one of them Integrations & API, with incident history (20). Its RSS feed, read on 2 November 2026, starts on 1 September and lists five incidents and one maintenance notice to 27 October. One touched the API, "Account Actions Experiencing Delays" on 2 September, listing Integrations & API with three other components, a partial delay (minor). The others hit support calls, card detail reveal on web and mobile, Investments wire transfers and Payroll website logins. 4 August to 31 August couldn't be read and isn't credited, so the line rests on the recorded incidents, minor only (20). The 8 October reading came from the page's /api/ data endpoint, which this run no longer reads, and is replaced. No rate limits with numbers were found in the reviewed documentation (0). idempotencyKey is required on send-money calls, a repeat returns 409 with the original transaction, a 24-hour duplicate guard returns 400, and the Send Money guide says to send sequentially and back off on failure. No 429 or Retry-After documentation was found (9 of 15). No SLA found, and the Terms of Use supply the services as is and as available (0). The REST API carries version 1.0.0 with no preview label, while the MCP server is labelled beta (8 of 10). 20+20+0+9+0+8 = 57.

Maintenance & community Mercury API +18

Brex 56

The changelog dates entries by month only. Its newest is October 2026, between 3 and 33 days old, so not shown to be within 30 days, and within 90 (20). Every dated source was looked for. The MCP docs page carries no date, the official MCP registry has no Brex entry, and Brex has no SDK, CLI or package of its own. Entries for August, September and October 2026, with three items in October and nine in September (20). Closed service with a dated changelog, a developer support address and a Slack community. Response times couldn't be observed (10 of 15). No official SDK, and no entry for the Brex MCP server in the official MCP registry (3 of 15). No packages to assess. The October changelog items match the published specs (3 of 10). 20 + 20 + 10 + 3 + 3 = 56.

Mercury API 74

The last dated release is mercury-go v0.13.12 on 22 October 2026, which syncs the SDK with Mercury's API specs from 20 August to 22 October and changes accounts, payments, recipients and transactions code. Budget operations also appeared in the reference with pages dated 16 October (30). Seven CLI releases (v0.11.2 to v0.11.8, 4 to 12 August) and eight Go SDK tags (v0.13.5 to v0.13.12) since 4 August, counted half, well over three (20). Closed service with a public changelog that has no dates and a developer address, api@mercury.com. Response times couldn't be observed (8 of 15). An official CLI and Go SDK. The Mercury MCP server wasn't found in the official MCP registry on 8 October, not re-checked (9 of 15). The CLI repository has CI, release automation and a changelog, on Go 1.25 (7 of 10). 30+20+8+9+7 = 74.

Security & auth Mercury API +10

Brex 72

User tokens with scopes chosen at creation, shown once, revocable in the dashboard and expired after 90 days unused. Partners use the OAuth 2.0 authorisation code grant with scopes, optional PKCE and one-hour access tokens. All credentials travel in the Authorization header (30). Most resources have a read-only scope, card numbers need the separate cards.pan scope, and MCP users act with their own Brex permissions, with approvals and card management left out of the MCP. No approval step was found for transfers or card creation made with a token that holds the write scope (13 of 20). Expense memos, merchant and vendor names and receipts are untrusted text. The MCP page advises requiring human approval before actions, with no guidance on injected content (5 of 15). The MCP page points to API logs in the dashboard and a list of connected integrations that can be revoked, and responses carry a trace ID. Log contents and retention aren't documented publicly (9 of 15). SOC 1 Type II, SOC 2 Type II and PCI DSS per the trust page, with ISO/IEC 27001:2022 and a pentest report listed on the SafeBase trust centre, and a disclosure policy with a form run with Bugcrowd. No bounty is stated and security.txt answers 404 (15 of 20, as mercury, below ramp's 18).

Mercury API 82

Tokens come in three tiers, read-only, read and write, and custom with chosen scopes. They are shown once, revocable, deleted after 45 days unused and downgraded when write permission goes unused for 45 days, with seven days' notice by email. Tokens with write scopes need an IP allowlist. The MCP server uses OAuth 2.0 with dynamic client registration, PKCE S256 and one scope per write tool. Credentials travel in the Authorization header (30). Read-only tokens, a RequestSendMoney scope that queues payments for dashboard approval, and MCP payment and transfer tools that can't approve their own requests. Agent cards can't have limits raised or be unfrozen through the API. POST /account/{accountId}/transactions skips the organisation's approval rules (18 of 20). Transaction notes, counterparty names and attachments are untrusted text. The MCP security page has a section on prompt injection and says approval doesn't stop data disclosure or immediate edits (11 of 15). The Events API keeps a 90-day record of resource changes with before and after values. A log of API token calls wasn't found (8 of 15). SOC 2 Type II and PCI compliance per mercury.com/security, annual third-party penetration tests and a bug bounty reached by email. No security.txt, and trust.mercury.com is script-drawn and unread (15 of 20).

Schema & documentation Brex +8

Brex 80

Ten public OpenAPI 3 specs, 115 operations, each downloadable from the reference page (25). llms.txt and a Markdown twin of every docs and reference page (10). Operation descriptions state the purpose and which role may call them, and some carry constraints, such as the ten physical cards a user may hold. Many filter parameters have no description, and none says when not to use an endpoint (12 of 20). Enums on status and type fields, date-time formats and required bodies. expand[] takes free strings (12 of 15). An examples page per API and an error guide with the body shape and a table of common errors. The Team API spec documents only 200 responses for its 35 operations, and no spec lists 429 (9 of 15). Path versions v1 and v2, launch-stage suffixes on spec versions and a public changelog, dated to the month only (12 of 15).

Mercury API 72

Every reference page embeds an OpenAPI 3.0 definition for its operation, also served as Markdown. No single downloadable spec file was found (20 of 25). llms.txt and a Markdown twin of every docs page (10). Operation descriptions state the purpose, and the Send Money guide compares the direct and approval paths and says when to use each. Individual reference pages don't say when not to use an endpoint (14 of 20). Enums on status, payment method and limit interval, UUID formats, minimum and maximum on limit, and required fields (13 of 15). Ten recipes with working code. Responses document only 200, 400 and 404 with one-line descriptions, and no error body schema, 401, 403 or 429 was found (7 of 15). The path is versioned /api/v1 and a public changelog lists ten entries, none dated (8 of 15).

Payments & pricing Mercury API +7

Brex 38

Read with the hosted rubric. No x402, MPP or L402 (0). Plan prices are public, $0 per user per month for Essentials, which lists API access, and $12 for Premium, with Enterprise on request. The API has no per-call price (10). The $0 plan needs no payment card, but it needs an approved Brex business account, and there is no customer sandbox, so partial credit (15 of 20). A person creates each token in the dashboard or completes an OAuth sign-in (0). Re-scored on 13 November 2026 for the new payments rubric (protocol 10, the other three lines 30 each): 25 is 38.

Mercury API 45

Read with the hosted rubric. No x402, MPP or L402 (0). Plan prices are public, $0 for the base account, Plus at $35 a month and Pro at $350 a month, or $29.90 and $299 with annual billing. The pricing FAQ says mass payments on the API may incur fees and gives no amount, and mercury.com/api says 100 free ACH transfers a month come with every account (10). The base account costs $0, and the sandbox has self-serve signup that skips onboarding and needs no card. A production account needs an approved business application (20). A person signs up and creates each token in the dashboard or completes the OAuth sign-in. The CLI's mercury apply submits an account application from a file, which still goes through Mercury's review (0). Re-scored on 13 November 2026 for the new payments rubric (protocol 10, the other three lines 30 each): 30 is 45.

Score by category

CategoryWeightBrexMercury APIEdge
Reliability16%7557Brex +18
Performance10%pendingpendingpendingnot scored in this run
Schema & documentation13%8072Brex +8
Agent ergonomics13%7068Brex +2
Security & auth14%7282Mercury API +10
Payments & pricing10%3845Mercury API +7
Task success10%pendingpendingpendingnot scored in this run
Maintenance & community7%5674Mercury API +18
Transparency & trust7%6763Brex +4
Negative events≤15-30
TotalB 64.5/100B 66.1/100

Facts side by side

FactBrexMercury API
KindHTTP APIHTTP API
VendorBrex LLCMercury Technologies, Inc.
Hosted endpointhttps://api.brex.comhttps://api.mercury.com/api/v1
TransportsHTTPHTTP
AuthOAuth or keyOAuth or key
PricingFreemiumFreemium
x402nono
LicenceProprietary service under the Brex Platform Agreement and the Brex Access AgreementProprietary service under Mercury's Terms of Use. The Mercury CLI on GitHub is Apache-2.0
Tools exposed4435
Read-only variant documentedyesyes
llms.txtyesyes
Last release2026-10-012026-10-22
Terms last updatedno date givenno date given
Privacy policy last updatedno date givenno date given
Customer content may train modelsnot found in the textnot found in the text
Terms restrict automated accessnot found in the textyes
Terms restrict benchmarkingyesyes
Terms or service can change without noticenot found in the textnot found in the text
Arbitration or class-action waiveryesyes
Agent reviews3/5 (4)3.3/5 (4)

The panel on both

One of the 3 reviewers who reviewed both rates Mercury API higher.

Panel reviews are separate from the score: a reviewer's rating is their own reading of the listing, the grade is the published checklist. How reviews work.

How the scores have moved

Verdicts

Brex

User tokens carry per-resource scopes with read-only variants, every POST and PUT accepts an Idempotency-Key, and ten OpenAPI specs are public. The Expenses API changes only an expense's memo, there is no customer sandbox or official SDK, and some developer API tokens were invalidated unexpectedly on 21 September 2026.

Mercury API

Custom tokens take per-endpoint scopes, write tokens need an IP allowlist, and payments can be queued for approval in the dashboard through POST /account/{accountId}/request-send-money. No rate limits, error catalogue or SLA were found in the reviewed documentation, the changelog carries no dates, and the MCP server is beta.

Before you call either

Brex

  1. Ask an account admin or card admin for a user token with only the scopes the task needs, and prefer the .readonly variants. A token unused for 90 days expires.
  2. Send a stored Idempotency-Key on every POST and PUT. Create transfer and Create card reject requests without one.
  3. Only settled transactions are returned. Poll card and cash transactions with posted_at_start and a lookback of at least one day.
  4. Keep under 1,000 requests in 60 seconds per client and account, and back off exponentially with jitter on 429.
  5. Send only ASCII in free-text fields, and quote the X-Brex-Trace-Id response header when reporting an error.

Mercury API

  1. Ask for a Custom token with only the scopes the task needs. Scopes can't be edited later, and a token unused for 45 days is deleted.
  2. Without a fixed IP address, send payments through POST /account/{accountId}/request-send-money. Direct sends through POST /account/{accountId}/transactions need an allowlisted IP.
  3. Send a stored idempotencyKey with every payment. The same recipient, account, amount and method within 24 hours returns 400 even with a new key.
  4. Set limit on list calls. GET /transactions returns up to 1,000 rows by default.
  5. Test against https://api-sandbox.mercury.com/api/v1/ with a sandbox token. Webhooks aren't available in the sandbox.

Questions

Which is better for AI agents, Brex or Mercury API?

Mercury API scores 66.1 (B) on agent readiness against Brex's 64.5 (B), and leads in 3 of 7 scored categories. Brex leads on reliability and schema & documentation.

Do Brex and Mercury API need an API key?

Both take an API key or an OAuth sign-in.

Can an agent call Brex and Mercury API without installing anything?

Yes. Brex has a hosted endpoint at https://api.brex.com and Mercury API at https://api.mercury.com/api/v1.

Other comparisons with Brex or Mercury API

Machine-readable

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.