Annalist · independent MCP server checks

com.modeltested/llm-benchmarks

Score 100 of 100 · issued 2026-10-21T01:15:39Z · valid until 2026-10-28T01:15:39Z · method mcp-1

+: answered 100% of our probes in 7 days

Model Tested. Independent test scores for open-weight LLMs: best model for your GPU or Mac, comparisons.

Endpoint https://modeltested.com/mcp (streamable-http) · source · listed in the Official MCP Registry

Probes in the last 7 days	3
Answered (%, 7 days)	100
Median connect time (ms)	1388
Sign-in	open
Tools	5
Tools changed in the last 14 days	no
Last tool change	–
Domain age (days)	–
Registry status	active

Verify it yourself. Ed25519 signature (base64url) over the RFC 8785 canonical JSON of `statement`, checked with the public key at https://api.annalist.dev/.well-known/annalist-key.json (key_id must match). Then check valid_until. The probes behind it are archived daily, and each day's archive hashes are in the public ledger.

Signed statement: https://api.annalist.dev/v1/verify/mcp?name=com.modeltested%2Fllm-benchmarks

Server authors: show this grade in your README. It updates daily.

[![Annalist MCP check](https://api.annalist.dev/v1/verify/mcp-badge.svg?name=com.modeltested%2Fllm-benchmarks)](https://annalist.dev/mcp/com.modeltested/llm-benchmarks)

Grades start at 100 and lose points only for measured problems, each listed above. The signature proves Annalist issued this statement unchanged; the facts rest on the measurements behind it, archived and hashed into the public ledger. Rankings are never for sale; Annalist's own service is checked by the same rules and left out of the leaderboards. Disputes: open an issue at github.com/ryanroskowski/annalist-ledger.