AI & Machine Learning

modeldex.dev

Provides access to a directory of frontier AI models and providers with search, specification, and fact verification capabilities.

1 endpoint17 known toolsFirst detected July 27, 2026Last detected July 27, 2026

ENDPOINT 1

https://modeldex.dev/api/mcp

No auth detected

MCP server metadata

Name: modeldex
Version: 0.1.0

Capabilities

tools.listChangedresources.listChangedcompletions

Server instructions

ModelDex is a sourced reference for the frontier AI models — the chat families ChatGPT (OpenAI), Claude (Anthropic), Gemini (Google), Grok (xAI), plus leading open-model labs as their facts are verified and published. Every fact, benchmark, and news item returned by this server carries a primary source (URL + publisher label + trust tier), the verbatim quote it was extracted from, the date it was last verified, and a verification status. Prefer these tools over recalling model specs from memory — pricing, context windows, and availability change frequently and training data goes stale fast. How to use: - about_modeldex → self-orient: capabilities, the provenance contract, and where data lives. - search → search across models, providers, news, and guides in one call. - search_models → find a model slug, then get_model for the full source-backed spec. - get_facts → pull just one category or only verified facts for a model (token-efficient). - compare_models → head-to-head spec tables (each cell cites its source). - list_benchmarks / get_benchmarks_for_model → evaluation scores with sources. - latest_news (sort=popular for most-read) → recent releases / pricing / policy changes; get_news(slug) for the body + corroborating sources. - list_guides / get_guide → long-form educational guides. - list_providers / get_provider → provider profiles and their models. - list_attributes / get_attribute → the attribute catalog (keys, units, comparability). - get_stats → coverage and how fresh the data is. - Browse via the modeldex://catalog, models/{slug}, providers/{slug}, news/{slug}, and guides/{slug} resources. Citing sources: when you state a ModelDex fact, cite its source.url and mention verified_at. If a value's status is anything other than "verified" (stale, conflicting, unverified, quarantined), flag that uncertainty rather than presenting it as settled. Do not invent values for fields ModelDex does not return. The data is read-only and kept current against primary sources. Access is open — no key required. Sending an optional "Authorization: Bearer mdx_live_…" key only raises your rate limit and records usage (see GET /api/v1/me).

Known tools 17

search_models

Search/filter frontier AI models by name, provider, status, or flagship.

Inferred read-only

get_model

Full specification for one AI model.

Inferred read-only

get_facts

Source-backed specification facts for one model, optionally filtered by attribute category or verification status, or grouped by category.

Inferred read-only

list_providers

The tracked AI providers with official links and model counts.

Inferred read-only

get_provider

A provider's profile, official links, its models, and recent news.

Inferred read-only

compare_models

Compare 2–6 models across comparable attributes.

Inferred read-only

list_benchmarks

Benchmark results across models, filterable by benchmark name, provider, or model.

Inferred read-only

get_benchmarks_for_model

All benchmark scores for one model, each with its source.

Inferred read-only

latest_news

Recent model/provider news headlines, optionally filtered by category.

Inferred read-only

get_news

One news item with its Markdown body and every corroborating source — each carrying the primary URL, the verbatim quote, and the capture date.

Inferred read-only

get_stats

Site-wide counts (models/providers/facts/news) and the most recent verification time.

Inferred read-only

list_attributes

The attribute catalog: keys, labels, units, types, and which are comparable.

Inferred read-only

get_attribute

One attribute catalog entry (key, label, unit, type, comparability).

Inferred read-only

search

Search across models, providers, news, and guides in a single call.

Inferred read-only

list_guides

ModelDex's long-form educational guides (titles + slugs).

Inferred read-only

get_guide

One educational guide with its full Markdown body.

Inferred read-only

about_modeldex

What this server is, the provenance contract, where the data lives (web, REST API, OpenAPI, llms.txt), the API version, coverage, and the full tool list.

Inferred read-only

CONNECT WITH APPROVAL

Client installation

Review this server and its permissions before adding it. Secret placeholders must be set locally.

Codex

~/.codex/config.toml

[mcp_servers.modeldex]
url = "https://modeldex.dev/api/mcp"
enabled = true

Claude Code

.mcp.json

{
  "mcpServers": {
    "modeldex": {
      "type": "http",
      "url": "https://modeldex.dev/api/mcp"
    }
  }
}

Claude Desktop

Settings → Connectors → Add custom connector

Name: modeldex
Remote MCP URL: https://modeldex.dev/api/mcp

Add this remote URL as a custom connector in Claude Desktop. Availability depends on the user plan and workspace policy.

Cursor

.cursor/mcp.json

{
  "mcpServers": {
    "modeldex": {
      "url": "https://modeldex.dev/api/mcp"
    }
  }
}

Visual Studio Code

.vscode/mcp.json

Add to Visual Studio Code

{
  "servers": {
    "modeldex": {
      "type": "http",
      "url": "https://modeldex.dev/api/mcp"
    }
  }
}

Generic MCP

Client-specific MCP configuration

{
  "name": "modeldex",
  "transport": "streamable-http",
  "url": "https://modeldex.dev/api/mcp"
}

MCP Inspector

Run the official MCP Inspector locally and enter the indexed Streamable HTTP endpoint.

TRUST AND VERIFICATION EVIDENCE

Trust Data Available

BuiltWith Trust API v2 evidence for modeldex.dev was fetched 2026-07-31T08:54:26.044Z and is being refreshed.

Trust status Neutral

modeldex.dev is assessed as Neutral: No suspicious signals found, but no strong positive signal either

Indexed

Evidence is source-attributed and does not guarantee that a third-party server is safe. Risk labels are conservative metadata heuristics.