Skip to content
MCP server

Clearbar model verdicts

Cheapest LLM that passes each task, with cost per successful task and pass rate

Status and history

Answered our latest tool-list check 10 hours ago with 2 tools. It will be marked verified after 3 successful checks in 7 days.

Install
Pricing
Free
Platforms
Web, API
Runs
Remotely
Version
1.1.0
Listing updated
3 days ago

About Clearbar model verdicts

Clearbar model verdicts is a free MCP server for ai models and llm tooling. It runs as a hosted remote server, so there is nothing to install locally.

Clearbar answers a narrow question: which model is the cheapest that still clears your quality bar for a given task. It counts every retry and failed attempt in the price, divides the total by tasks that passed, and shows the retry tax separately. Calls go through OpenRouter, and the answer is dated by an OpenRouter price snapshot.

The public verdicts run the same method on open datasets: Amazon Polarity sentiment, glaive function-calling, SMS spam, BANKING77 intents and JSON extraction, each with a pass-rate bar and a downloadable script to repeat the run. A paired test compares a candidate with the model you use today, within a stated margin of 5 points by default. Status chips mark each model as cleared, borderline, below bar or errored using a 95% Wilson interval.

The MCP server is read-only and needs no account or key. Two tools let a coding agent list verdicts or fetch the full table for one task, and every answer links to running the same verdict on your own examples.

Install

Run in a terminal

claude mcp add --transport http clearbar https://clearbar.grain64.com/mcp

Add to Settings → Connectors → Add custom connector

https://clearbar.grain64.com/mcp
Install in CursorOpens Cursor, which asks you to confirm. Or add it by hand:

Add to ~/.cursor/mcp.json

{
  "mcpServers": {
    "clearbar": {
      "url": "https://clearbar.grain64.com/mcp"
    }
  }
}
Install in VS CodeOpens VS Code, which asks you to confirm. Or add it by hand:

Add to .vscode/mcp.json

{
  "servers": {
    "clearbar": {
      "type": "http",
      "url": "https://clearbar.grain64.com/mcp"
    }
  }
}

Add to ~/.codeium/windsurf/mcp_config.json

{
  "mcpServers": {
    "clearbar": {
      "serverUrl": "https://clearbar.grain64.com/mcp"
    }
  }
}

Add to your client's MCP settings (most use this shape)

{
  "mcpServers": {
    "clearbar": {
      "type": "http",
      "url": "https://clearbar.grain64.com/mcp"
    }
  }
}

Tools this server provides (2)

Read directly from the server.

list_verdicts
Which LLM is the cheapest that still passes? Lists every published Clearbar verdict at once: the task, the pass bar, the cheapest passing model, its cost per successful task (retries and failures counted) and the page to check it. Optional filter by task type. Next call: get_verdict with {"task":"spam detection"} (or a slug from this list) for the full table.
get_verdict
Answers: which is the cheapest LLM that still passes for <task>? Give the task in plain words (for example "spam detection", "sentiment", "json extraction", "tool calling", "intent classification") or a verdict slug, and get the cheapest model that clears the bar, its pass rate and cost per successful task in US dollars, the baseline it is compared with, every candidate's status, plus links to the

What it does

Cost per success
Retries and failures are priced in and divided by passing tasks.
Paired comparison
Tests whether a cheaper model is no more than a set margin worse.
Public dataset verdicts
Five tasks from open datasets, each with its bar and price date.
Re-runnable scripts
Download the script that produced a verdict and repeat it.
Judge calibration
For free-text tasks, label answers yourself to see how the judge agrees.

Good for

  • Ask a coding agent which model is cheapest for spam detection
  • Check whether a cheaper model can replace your current one for classification
  • Compare cost per successful tool call across models
  • Re-check a verdict after prices change
  • Choose a model for JSON extraction from messy documents

Pricing

Free

Free during early access: 3 free runs a month; model calls use your own OpenRouter key

Prices change. Check clearbar.grain64.com before you buy.

Clearbar model verdicts FAQ

Do I need an account for the MCP server?

No. It is open to every client with no account and no key, and is read-only.

How is cost per successful task computed?

Usage cost from OpenRouter is summed across all attempts, including retries, then divided by tasks that passed.

How repeatable are results?

The page says a re-run lands within about 5 points of pass rate and 10 percent of cost, because models are not fully deterministic at temperature 0.

What does running my own examples cost?

The pricing page lists 3 free runs a month during early access, with model calls billed to your own OpenRouter key.

Is Clearbar model verdicts free?

Yes. Clearbar model verdicts is free to use.

How do I add Clearbar model verdicts to Claude Code?

Run this command in your terminal: claude mcp add --transport http clearbar https://clearbar.grain64.com/mcp

What are the best alternatives to Clearbar model verdicts?

The closest alternatives to Clearbar model verdicts are Claude Code Ultimate Guide, signafAIde, Toolkien, Lugh and HypePaper.

Alternatives to Clearbar model verdicts

  1. MCP server for searching the Claude Code Ultimate Guide from any MCP client

    Runs locally
    • Open source
    • 6.5K GitHub stars
    • AI models and LLM tooling
  2. signafAIde

    MCP server

    Local MCP server to set up, verify and check coverage of signed agent action records

    Runs locally
    • 5 tools
    • Freemium
    • Security
  3. Toolkien

    MCP server

    One MCP gateway for all your MCP servers, with code mode, permissions and audit

    Not checked yet
    • 3 tools
    • Remote
    • Open source
    • AI models and LLM tooling
  4. Lugh

    MCP server

    Write, check and dry-run Lugh agent programs with policy, budgets and approvals

    Runs locally
    • 5 tools
    • Open source
    • AI models and LLM tooling
  5. HypePaper

    MCP server

    Research papers, benchmarks, citations and topic experts for AI assistants

    • 58 tools
    • Remote
    • Freemium
    • AI models and LLM tooling