What this page covers
- Facts checked
- Sources
- 25 listed, with access dates
- Screenshots
- None on this page
- Score
- No numeric score on this page
Documentation-only comparison. An AI-assisted research pass on 7 Nov 2026 read both vendors' homepages with FAQs expanded, both pricing pages, both terms of service in full, Monid's full docs index plus its pricing, x402 and integration pages, both READMEs, the GitHub API, the YC directory, investor pages, a PitchBook public profile, dated press and two @MonidHQ posts viewed in a browser (the 16 Oct 2026 Pakistan-time 0% markup announcement and its free-credits reply). No accounts, keys, calls or spend. No score or winner is published because the rubric criteria need tests. No vendor logos or page captures are published. AI-agent setup lines on vendor pages were excluded as untrusted content and not followed.
What they share and where they differ
treg and Monid sell the same basic product. An AI agent gets one key and one prepaid balance for thousands of paid APIs without a subscription to each vendor. Both describe themselves as an OpenRouter for agent tools [2] [7].
Several apparent differences are shared. Both serve a catalog on their own provider keys, searchable by job, priced before the call [2] [9] [23]. Both price some endpoints per call and others per result or per success [9] [23]. Both give new accounts $1 of credit [6] [23]. Both answer an empty balance with HTTP 402 [2] [11]. Both document usage limits: treg sets per-member daily call caps in its terms [3], and Monid documents workspace budgets and per-run caps [11].
The documented differences are narrower. treg also runs your team's own provider keys and connected accounts, such as Search Console, Google Analytics, Google Ads and Business Profile, and never meters those calls [23], except "X through treg's own X app, which X bills per use" [4]. treg's search shows providers "side by side with prices" and "choosing is yours", apart from opt-in routed endpoints "where treg picks the provider for you" [2], with success rates and speed it says it measured [1]. Monid's discover ranks candidates itself, with automatic failover at the gateway [7] [8]. Monid also takes per-run payment in USDC through x402 [10]; treg says top-ups are by card today [4].
Scorecard: why there is no score
Our rubric leaves any untested criterion unscored. On the check date that was almost all of them, so this page publishes no score and no winner.
| Pillar | Status | Evidence a score would need |
|---|---|---|
| UI and UX | Not scored | Logged-in sessions on both, same day, same screen size |
| Capability for the job | Coverage documented; quality and reliability not scored | The same keyword and enrichment job run on both |
| Cost and pricing | Pricing documented; cost for a defined job not scored | A metered run under a spend cap, with billed amounts compared |
| Technical fit | Docs read; auth, limits and errors not exercised | API and MCP calls with timings and error cases |
| Vendor durability | Company facts reported below, not scored | Support replies and changelog activity over time |
What each one is
treg is an open-source tool registry and paid API gateway hosted at treg.to. Its terms say Superdesign Dev Pty Ltd operates the hosted service [3]. It serves a catalog of external endpoints on its own provider keys, and it holds a team's own keys, OAuth connections, CLIs and skills so each teammate's agent can use them without seeing the secret [2]. On the check date, treg.io returned the same page title as treg.to [5].
Monid is a tool gateway and marketplace for AI agents. Its site footer reads "Monid Inc · Delaware C-Corp · San Francisco" [6]. An agent searches the catalog (discover), reads an endpoint's schema and price (inspect), then calls it (run). The public repository holds the connector layer that describes each provider and endpoint, under the MIT licence [8].
GitHub records the treg repository as created on 15 July 2026 and Monid's on 26 August 2026 [17] [18].
How this comparison was researched
- An AI-assisted research pass read, on 7 November 2026, both homepages with FAQs expanded, both pricing pages, both terms in full, Monid's docs index and its pricing, x402 and integration pages, both READMEs, the GitHub API, the Y Combinator directory, investor pages, PitchBook and dated press [1] [2] [3] [6] [9] [10] [11] [23] [26].
- X posts were checked by opening each post and its thread in a browser [19] [21].
- No product accounts, API keys, calls or spend were used. Sign-up, dashboards, speed, output quality, errors and support were not tested.
- Setup lines on both sites that tell AI agents to fetch a file were treated as page content; nothing was fetched or installed.
- treg's own treg vs Monid page [4] is used only for treg's statements about itself, labelled as vendor claims.
Design and layout: not rated
The rubric rates design and onboarding from logged-in sessions and timed page loads, so neither product is rated. No vendor images are used in this text-only review.
Keys, accounts, sign-in and roles
| Question | treg | Monid |
|---|---|---|
| Signing in | An email address, "either by one-time code or through a supported identity provider" [3]. The CLI defaults to GitHub, with an email code or a token for agents and CI [2] | Accounts live in the app.monid.ai dashboard [7]. Each API key "is bound to the workspace it was created in" and carries scopes [11]. Sign-in methods are not described in the pages read |
| Provider keys on the hosted service | "Your own key always wins over treg's, and those calls are never metered." [2] | FAQ: "Monid holds the keys and brokers calls on your agent's behalf." [6] A bring-your-own-key option for the hosted service is not listed in the docs index [11]. The open-source engine runs locally with your own vendor key [8] |
| Your own accounts by OAuth | The terms list Google Search Console, Google Analytics, Google Business Profile, Google Ads, YouTube, LinkedIn, Slack and X [3]. The pricing FAQ says connected accounts are never metered [23]. treg's comparison page names one exception: "X through treg's own X app, which X bills per use" [4] | Not listed as a feature in the docs index [11]. The terms carry a general clause on linking third-party accounts [22] |
| Serving your own customers | A "platform customer" path: your product tags calls with X-Treg-Meta and invoices from usage/by-tag [3]. The README describes agents pinned to one customer's records [2] | Three modes: call Monid directly so "Users never see Monid", let users "sign in with Monid and act on their own workspace", or run your own proxy to "Control validation, billing, and endpoint access" [26]. OAuth clients come from Monid support [26] |
| Team roles | Owner, admin, member and viewer, with tool access and a daily call cap per member [2] [3]. Team owners can turn on enterprise SSO through WorkOS [2] | API key scopes such as runs:create [11]. Member roles are not described in the docs index [11] |
OAuth means different things here: treg's connects your own Google, ad and social accounts, while Monid's lets your product's users sign in with their own Monid workspace.
Pricing, billing units, misses and errors
| Question | treg | Monid |
|---|---|---|
| Billing unit | "Metered per catalog call (or per success where the provider bills that way)" [23]. Catalog rates appear per call, per success and per result, for example people data "from $0.0001/success" [23]; homepage examples are priced per keyword and per row [1] | Per call, or per result "with an optional flat base fee" [9]. The README adds "a rate per unit such as a thousand characters or a second of video" [8], llms.txt mentions "units such as resolution and duration" [7], and the x402 guide names "Metered endpoints (time-billed)" [10]. Owned resources such as phone numbers are paid "over time" [11] |
| Misses and failed calls | "On per-success tools, a miss settles at $0.00. On per-call tools, every call costs regardless of what comes back." [23] | FAQ: "We only charge when the request actually delivers results to your agent. If you think you were charged in error, reach out and we'll refund it." [6] The README says an unmatched company or unresolved person "settles at zero" [8]. A per-call price is "A flat fee per execution, regardless of how many results are returned" [9] |
| Other no-charge cases | Responses over 8 MiB that need settlement "return a 502 without charging"; an endpoint with no published price is refused [2] | x402 timeouts are "Zero-billed" [10]. "A stopped run is not charged", and a dispute command covers "runs that returned broken or useless results" [11] |
| Balance runs out | HTTP 402 insufficient_balance with balance_micro, estimated_cost_micro and a topup_url [2]. Own-key calls keep working [23] | A run can fail with WALLET_INSUFFICIENT_FUNDS (402); the checkout link goes to a person because "An agent cannot pay by itself" [11] |
| Limits | Per-member daily call caps, in the terms [3]. treg's comparison page also claims "A team daily spend cap, a per-call cost ceiling, per-member daily call caps and daily budgets per tagged customer (advisory)" [4] | Workspace budget controls and per-run caps. "A run blocked by a workspace budget or run cap returns 200 with status BLOCKED." [11] |
| Markup | Terms: "we add no markup" [3]. Homepage: "we earn on volume pricing with vendors - not on you" [1] | Terms: endpoints labelled "0 Markup" carry no platform fee, and "Endpoints without the label may include a platform fee." They add "Monid may obtain data on different terms under its own agreements with providers." [22] |
| Starting credit | "$1.00 free once" for new verified accounts that create an eligible team [2] | "Start with $1 in free credit." [6], with no card required [7]; one signup credit per workspace [11]. An @MonidHQ post on 16 Oct 2026 (Pakistan time) promoted "free credits" [19] |
| Payment | Card top-ups; x402 or stablecoin payment "Not yet", per treg's comparison page [4]. Automatic top-ups happen only if you turn them on [3] | Terms list Visa, Mastercard, American Express, PayPal and Discover [22]. x402 in USDC on Base or Monad with a $0.01 floor, not for time-billed endpoints; discovery and inspection still need an API key [10] |
| Refunds | Referral credit is "never redeemable for cash and never refundable" [3]. No general refund clause in the terms read | The terms point to a "Return Policy posted on the Services" [22] that the sitemap did not show [20] |
| Vendor example prices | Keyword volume $0.0018 per keyword; backlinks $0.00018 per row; funding data $0.01 per call [1] | Twitter search $0.003 per call; reviews $0.001 per result plus $0.002 per call [9] |
The written miss policies differ for per-call tools: treg charges every call, while Monid says failed or unmatched requests cost nothing. Neither was tested.
Monid's fee wording has changed. Forkast on 8 Oct 2026 described its pricing as "roughly $0.0013 per call plus a 10% platform fee" [14]. On 16 Oct 2026 (Pakistan time) @MonidHQ posted "Announcing: Monid is now 0% markup. No subscriptions. No markups." [21]. On 7 Nov its homepage FAQ said "Most endpoints carry 0 Markup" [6], and its terms, last updated 16 Oct 2026, make the label the test [22]. No cost for a defined job is given for either product.
Terms: embedding, resale, agents and liability
| Question | treg | Monid |
|---|---|---|
| Operator and law | Superdesign Dev Pty Ltd; Delaware law, Delaware courts [3] | Monid, Inc., registered in Delaware; California law, with binding AAA arbitration in San Francisco [22] |
| Use inside a product you sell | Allowed for "Using the hosted treg.to API inside your own product" when done "with your own billing layer and the pass-through attribution mechanics", which the terms call "calling our API, not reselling it". A product that "uses X-Treg-Meta to tag calls by customer, and invoices from usage/by-tag" is a platform customer and "no special agreement is needed" [3] | The docs describe direct, OAuth and proxy integration [26]. The terms say "The Services may not be used in connection with any commercial endeavors except those that are specifically endorsed or approved by us", and grant access "solely for your personal, non-commercial use or internal business purpose" [22] |
| Resellers | The restriction covers redistributing the open-source software "as a competing hosted, managed, or embedded registry service without prior written authorization" [3] | "We reserve the right to limit or prohibit orders that, in our sole judgment, appear to be placed by dealers, resellers, or distributors." Prohibited activities include use "for any revenue-generating endeavor or commercial enterprise" [22] |
| Agents and automation | "An agent acting under your token is you, as far as these Terms are concerned." [3] | Users represent they "will not access the Services through automated or non-human means, whether through a bot, script or otherwise", and prohibited activities include "Use a buying agent or purchasing agent to make purchases on the Services." [22] |
| Upstream provider terms | For connected accounts and "any API you register yourself", calls "are subject to that provider's terms as well as these" (section 05). Section 07 bars breaking "the terms, rate limits, or anti-abuse measures of any upstream API" [3] | Your relationship with providers behind linked third-party accounts "IS GOVERNED SOLELY BY YOUR AGREEMENT(S)" with them (section 11) [22] |
| Code licence | Apache 2.0 "plus additional terms" [3]; the homepage line reads "100% open source · Apache 2.0" [1] | Connector layer under MIT; the README does not say the hosted platform code is public [8] |
| Availability and liability | "Free, and offered as-is"; "no uptime commitment and no support commitment"; liability capped at the greater of 12 months' fees or US$100, and "While the Service is free, that means US$100", in the same terms that describe metered billing [3] | "ON AN AS-IS AND AS-AVAILABLE BASIS"; "We cannot guarantee the Services will be available at all times"; liability "LIMITED TO THE AMOUNT PAID, IF ANY" [22] |
Monid's docs are written for AI agents and describe embedding [7] [26], while its terms restrict commercial use without approval, reseller orders, buying agents and automated access. The pages read do not say how Monid applies those clauses to agent use, so ask Monid in writing before building a paid product on it. treg's permission depends on your own billing layer and its attribution mechanics.
Neither set of terms, as read, grants a right to resell the data providers return, and paying for API access does not create one. Check with the vendor and the upstream provider. This is a reading of public terms, not legal advice.
Routing, connections and catalog size
- Ways to connect. treg offers a CLI, an HTTP API with OpenAPI docs, MCP endpoints and a Claude Code plugin; each call carries an X-Treg-Token header [2]. Monid offers a skill, a remote MCP server, a CLI and a REST API [7].
- Provider choice. treg "does not silently pick or fail over between providers for you", except on opt-in routed endpoints, and may serve the same endpoint through "a treg-owned relay account, disclosed on the response; a team can opt out" [2]. Monid's discover ranks the catalog by the job and returns each candidate's price, live health and observed latency [8]; Monid documents "pooled rate limits and automatic failover" [7]; its FAQ says "Your agent retries on the next-best provider when one is throttled" [6].
- Writes into apps. treg lists social posting and ad management through connected accounts [1] [4]. Monid is "focused on reading and fetching data, not app actions" [7], though it lists agent email [6] and agentic phone calls [7].
| Source | Date | Published figure |
|---|---|---|
| treg homepage [1] | 7 Nov 2026 | "4,100+ endpoints", 126 providers |
| Monid homepage [6] | 7 Nov 2026 | Badge "2900+ TOOLS"; stat "2900+ Tools across 97+ providers"; footer "2,985 tools & APIs" (read about 15:45 UTC) |
| Monid llms.txt [7] | 7 Nov 2026 | "2,800+ tools" |
| Monid README [8] | 7 Nov 2026 | "2,000+ tools across 72+ providers" |
| @MonidHQ post [21] | 16 Oct 2026 (Pakistan time) | "2,500+ tools" |
Company and operations
| Field | treg (Superdesign) | Monid |
|---|---|---|
| Operator | Superdesign Dev Pty Ltd, per treg's terms [3] | Monid, Inc., registered in Delaware, per its terms [22]; the site footer adds San Francisco [6] |
| Founders | treg's pages name no founder. Capital Brief reported on 21 Oct 2026 that Jason Zhou started Superdesign after leaving Relevance AI [15] | Feiyou Guo and Shengkun Ye, per investor Founders, Inc. [13] |
| Funding | PitchBook's public profile for Superdesign (website www.superdesign.dev, Sydney) lists a completed "Early Stage VC" deal dated 01-Aug-2026, with no amount shown and investor Side Stage Ventures [24]. Side Stage Ventures lists Superdesign in its portfolio [25]. It is a third-party entry the company has not confirmed, covering Superdesign as a whole, not treg alone; treg's pages announce no round | Forkast reported on 8 Oct 2026 a $7.7M seed announced 6 Oct and led by Long Journey Ventures, after a $2.1M pre-seed in September: $9.8M disclosed [14]. Monid's own sitemap lists no funding post (checked 7 Nov 2026) [20] |
| Y Combinator | No directory entry for treg or Superdesign [16] | No directory entry for Monid [16] |
| Revenue figures | No ARR or MRR figure in the sources read. PitchBook lists the 01-Aug-2026 deal's stage as "Generating Revenue"; that describes the deal row, and its current-revenue field is blank [24] | No ARR or MRR figure in the sources read. Forkast cited "more than four million agent transactions" and called them company-reported [14] |
| Team size | Superdesign had six people, per Capital Brief on 21 Oct 2026 [15]; treg's share not stated | Founders, Inc. lists a team of 2 [13] |
| Repository activity | About 4,404 commits on the default branch, 7,428 stars, last push 7 Nov (snapshot about 08:37 UTC, 7 Nov 2026) [17] | 59 commits, 6,139 stars, last push 3 Nov (same snapshot) [18] |
Commit counts do not compare: treg's repository holds the whole registry and Monid's the connector layer [8]. Capital Brief's figures cover Superdesign as a company, including its design product.
Trade-offs on paper
treg. Strongest on paper when your own keys and connected accounts carry the work, since treg says those calls are never metered, X excepted [2] [4] [23], and when you want embedding permission written into the terms, with its conditions [3]. Its gaps: per-call tools charge misses [23], there is no automatic failover outside routed endpoints [2], top-ups are card only [4], and the terms still say "Free, and offered as-is" and cap liability "While the Service is free" beside metered billing [3].
Monid. Strongest on paper when you want the gateway to rank candidates and fail over providers [7] [8], or to pay a run in USDC [10]. Its gaps: catalog totals differ across its own pages [6] [7] [8], zero markup covers labelled endpoints only, and its terms on commercial use, resellers, buying agents and automated access sit beside agent-focused docs without a stated reconciliation [22].
Both. Neither promises availability, and both cap liability at modest amounts [3] [22].
Which fits which job
- If the agent must work inside your own Search Console, Google Analytics, ad or social accounts, treg documents that path, unmetered except X, which X bills per use [3] [4] [23]. Monid's docs index does not list it [11].
- If you already pay for provider keys and want them used first, treg documents that for its hosted service [2]. The Monid pages read describe own keys only for its open-source engine run locally [8].
- If you want the gateway to rank candidates and fail over between providers automatically, Monid documents that [7] [8]. treg shows priced options side by side for you to choose, plus opt-in routed endpoints and a web-access skill that falls back "to the next vendor when one fails" [2].
- If the agent should pay per run in USDC, Monid documents x402 [10]; treg says not yet [4].
- If you plan to embed either in a product you sell, read both terms first. treg allows it with your own billing layer and attribution [3]; Monid's terms point to its approval [22].
There is no overall winner.
FAQ
Are treg and treg.to the same thing?
Yes. treg is the product and treg.to hosts it [2]. treg.io returned the same page title on the check date [5].
Is either one free?
Both start with $1 of credit, then bill per use from a prepaid balance [2] [6]. On treg, calls on your own keys and connected accounts are not metered [23], except X, which X bills per use [4].
Do they add a markup?
treg's terms say "we add no markup" [3]. Monid's terms say labelled "0 Markup" endpoints carry no platform fee and unlabelled ones may [22].
Do failed calls cost money?
On treg, a per-success miss settles at $0.00 and a per-call tool charges every call [23]. Monid's FAQ says it charges only when a request "actually delivers results" [6]. Neither was tested here.
Can I build either into a product I sell?
treg's terms allow it with your own billing layer and attribution mechanics [3]. Monid's terms restrict commercial use unless Monid approves it [22]. API access does not give resale rights to provider data.
Are they open source?
treg's registry is public under Apache 2.0 with added terms [3]; Monid's connector layer is MIT [8].
Sources, disclosure and changelog
Method note: this page comes from an AI-assisted research pass that read public documents in a browser. It includes no product accounts, API keys, calls or tests.
Changelog: first version, built from public documents on the check date in the byline. Prices, catalog counts and terms change, so check each vendor's live pages before you decide. The source table below lists every source with its access date.