Datasets:
license: cc-by-4.0
pretty_name: Anchor Terminal grades
language:
- en
tags:
- ai-agents
- mcp
- benchmark
- tools
- open-weights
size_categories:
- 1K<n<10K
configs:
- config_name: latest
default: true
data_files: grades.jsonl
- config_name: history
data_files: history/*.jsonl
Anchor Terminal grades
Every tool AI agents run on, graded from public evidence against a published checklist, one row per listing. Model APIs, open-weight models, agent frameworks and harnesses, MCP servers, HTTP APIs, data providers, scrapers and payment protocols. The grades are Anchor Terminal's, the method is at https://www.anchorterminal.com/benchmark/, and the same data is on every listing page, in the JSON API and over MCP.
Updated 2026-11-13. 1701 listings, 1524 ranked, 229 agent-ready (BB or better). Grades AA 3, A 25, BB 201, B 391, C 457, D 287, E 165, F 172. Methodology v0.7, October 2026 research run. The latest config is today's grades; history holds one file for each day the grades changed, so a listing's score over time is its rows across those files.
What a row is
A listing graded by one of two checklists. checklist says which. tools (1541 rows) scores reliability, schema and documentation, agent ergonomics, security and auth, payments and pricing, maintenance and community, and transparency and trust, with performance and task success still pending until the probes and task suites run. open-weights (160 rows) scores capability, reach, risk, agent interface and upkeep. A pending or unscored category is null. score is the weighted total over the assessed categories, 0 to 100, after negative points for dated negative events, and grade is the band it falls in. rank is the listing's place among all ranked listings, and 0 when it isn't ranked, which is the case for protocols, shut-down listings and open-weight models (those rank among themselves, by category_rank).
review_count and avg_rating are the panel's desk reviews, written from public material with no calls made, and they never move a score. own marks one of our own products, graded under the rule on https://www.anchorterminal.com/benchmark/#own. competes_with names our product a listing competes with. vendor_linked says the vendor has verified the listing with a link back, which doesn't affect the grade.
Columns
| column | type | what |
|---|---|---|
slug |
string | the listing's id, the last part of its page address |
name |
string | the listing's name |
vendor |
string | who makes it |
kind |
string | mcp, http-api, sdk, model, router, platform, weights, framework, harness or protocol |
area |
string | the area of the directory |
category |
string | the category slug |
checklist |
string | tools or open-weights |
grade |
string | AA, A, BB, B, C, D, E or F |
score |
float | 0 to 100 |
agent_ready |
bool | grade BB or better |
ranked |
bool | whether rank is set |
rank |
int | place among ranked listings, 0 when not ranked |
rank_of |
int | how many listings are ranked |
category_rank |
int | place in its category |
confidence |
string | high, medium or low, how much of the checklist the evidence covered |
assessed |
date | when the scores were last checked |
methodology |
string | the checklist version |
run |
date | the research run |
negative |
int | points taken off for dated negative events |
score_* |
int or null | per-category scores, 0 to 100, null when pending or not on this checklist |
review_count |
int | panel reviews |
avg_rating |
float | panel average out of 5, 0 with no reviews |
pricing |
string | free, freemium, usage, byo-plan or paid |
price_summary |
string | one short price |
auth |
string | none, api-key, oauth, pat or mixed |
where |
string | hosted, local, both, library or spec |
x402 |
string | yes, partial or no |
capabilities |
string | capability ids, space-separated |
own |
bool | one of our own products |
competes_with |
string | the slug of our product it competes with, or empty |
disclosure |
bool | the listing carries a disclosure |
retired |
string | the date it was retired, or empty |
github_stars |
int | stars on its repository, 0 when it has none |
npm_weekly |
int | weekly npm downloads |
pypi_weekly |
int | weekly PyPI downloads |
vendor_linked |
bool | the vendor has verified the listing |
url |
string | the listing page |
markdown_url |
string | the listing's Markdown twin |
json_url |
string | the listing's full JSON record |
as_of |
date | the day the row was published |
Licence and attribution
CC BY 4.0. Credit Anchor Terminal and link the listing (url) or https://www.anchorterminal.com when you pass a grade on. The terms are at https://www.anchorterminal.com/terms/#data.
Anchor Terminal grades, https://huggingface.co/datasets/AnchorTerminal/grades, CC BY 4.0. Anchor Terminal (https://www.anchorterminal.com).
Where the rest is
- Every listing in full, with the reason and sources for each score, the prices and how to connect: https://www.anchorterminal.com/api/v1/tools/{slug}.json, or the listing page and its Markdown twin (
url,markdown_url) - The method, weights and grade bands: https://www.anchorterminal.com/benchmark/
- The MCP server, no key: https://www.anchorterminal.com/mcp
- A grade over DNS, for agents with no HTTP:
dig TXT <slug>.grade.anchorterminal.com - Disputes and corrections: https://www.anchorterminal.com/builders/#disputes
What isn't here yet. Performance and task success, which wait for measurement, and the reviews themselves, which are on each listing.