Seb Labs
Typed probabilities for agent decisions: route, classify, screen evidence, label in bulk
- Pricing
- Free trial
- Platforms
- API, CLI
- Runs
- Remotely
- Version
- 1.1.0
- Listing updated
- 2 days ago
About Seb Labs
Seb Labs is an MCP server for ai models and llm tooling. It runs as a hosted remote server, so there is nothing to install locally.
Instead of generating prose, the service returns a probability for every option you supply and names the most likely one. Probabilities sum to one, options keep your order, and a state flag turns to 4 when the top two are within 0.1 of each other, a signal that a human or more context is warranted.
Live inference runs a Seb-9B-v2-NVFP4 model on the vendor's own GPUs. Requests can be routing questions with 2 to 20 options, or typed questions of kind choice, noul or score for rubric scoring and evidence screening. Limits include a 32 KB body, 12,000 characters of context, a 15-second deadline and 80 requests per second. Failed or invalid inference is not charged.
The docs state the tools do not trade, move robots or call downstream targets. New accounts get $5 in credit with no card, and a free demo returns fixed illustrative distributions without an account. A finance model is in closed beta.
Install
Run in a terminal
claude mcp add --transport http seblabs https://mcp.seblabs.dev/mcpAdd to Settings → Connectors → Add custom connector
https://mcp.seblabs.dev/mcpAdd to ~/.cursor/mcp.json
{
"mcpServers": {
"seblabs": {
"url": "https://mcp.seblabs.dev/mcp"
}
}
}Add to .vscode/mcp.json
{
"servers": {
"seblabs": {
"type": "http",
"url": "https://mcp.seblabs.dev/mcp"
}
}
}Add to ~/.codeium/windsurf/mcp_config.json
{
"mcpServers": {
"seblabs": {
"serverUrl": "https://mcp.seblabs.dev/mcp"
}
}
}Add to your client's MCP settings (most use this shape)
{
"mcpServers": {
"seblabs": {
"type": "http",
"url": "https://mcp.seblabs.dev/mcp"
}
}
}Tools this server provides (5)
Taken from the project documentation; the running server may differ.
- seb_demo
- Inspects one of four illustrative presets, free.
- seb_decide
- Submits a text state with 1 to 10 typed questions.
- seb_route
- Classifies context among 2 to 20 caller-provided options.
- seb_batch_decide
- Processes up to 100 states with shared questions.
- seb_fin_analyze
- Runs the Seb-9B-Fin model for approved beta accounts.
What it does
- Probability per option
- Returns a distribution rather than a single label.
- Close-call flag
- A state value marks near-ties worth handing off.
- Typed questions
- Choice, noul and score questions with instructions and criteria.
- Batch labelling
- Up to 100 states with shared questions in one call.
- Free demo
- Four preset scenarios work without a token.
- Plugin for agents
- A skill for Claude Code and Codex explains when to use it.
Good for
- Route support tickets to the right team
- Screen whether evidence supports a claim
- Score text against a rubric
- Label large collections of text in batches
Pricing
Free trial
$5 starting credit with no card; usage-based afterwards
Prices change. Check seblabs.dev before you buy.
Seb Labs FAQ
How is it billed?
New accounts receive $5 in credit with no card. Failed or invalid inference costs nothing.
What are the request limits?
32 KB body, 12,000 context characters, 2 to 20 options, a 15-second deadline and 80 requests or 100,000 input tokens per rolling second.
Can I try it without an account?
Yes, seb_demo and the preset requests return illustrative results for free.
Does it act on its decisions?
No. The docs say tools do not execute trades, actuate robots or call downstream targets.
Is Seb Labs free?
Seb Labs is a paid product with a free trial.
How do I add Seb Labs to Claude Code?
Run this command in your terminal: claude mcp add --transport http seblabs https://mcp.seblabs.dev/mcp
What are the best alternatives to Seb Labs?
The closest alternatives to Seb Labs are Claude Code Ultimate Guide, signafAIde, Toolkien, Lugh and HypePaper.
Alternatives to Seb Labs

Claude Code Ultimate Guide
MCP serverMCP 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

signafAIde
MCP serverLocal MCP server to set up, verify and check coverage of signed agent action records
Runs locally- 5 tools
- Freemium
- Security

Toolkien
MCP serverOne 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
Lugh
MCP serverWrite, check and dry-run Lugh agent programs with policy, budgets and approvals
Runs locally- 5 tools
- Open source
- AI models and LLM tooling
HypePaper
MCP serverResearch papers, benchmarks, citations and topic experts for AI assistants
- 58 tools
- Remote
- Freemium
- AI models and LLM tooling

One compact entry point that finds, describes and runs hundreds of pay-per-call data tools
- 3 tools
- Remote
- Freemium
- Developer tools