Free daily comparisons of AI models on business jobs, measured in right answers per dollar.
Maintain this server? Claim it
Call this server's tools from your browser. Nothing to install.
Requests go through mcptry because browsers can't call most MCP servers directly. Header values are sent to the server for that request only and are never stored or logged.
Tool list from our last check. Connect live to refresh it.
Use to compare the models and systems on one job on one day. Leave date out for the newest day. Returns every model's and system's published result that day, NL;NL Labs' own models and systems marked. Extract's documents days also carry invented, missed and unreadable counts. Fields: cost_per_1000_right_usd and ipd are exact; cost_shown and ipd_shown are how the page shows them. assembled_by_us is 1 when the answer was assembled by us, else 0. A count marked "not published for this model" is not published for that model.
Job id (see list_jobs).
YYYY-MM-DD. Default: newest day.
Pick your client. Values in capitals, like YOUR_TOKEN, are yours to fill in.
Run in a terminal
claude mcp add --transport http results https://mcp.nlnllabs.com/mcp
Use to compare the models and systems on one job on one day. Leave date out for the newest day. Returns every model's and system's published result that day, NL;NL Labs' own models and systems marked. Extract's documents days also carry invented, missed and unreadable counts. Fields: cost_per_1000_right_usd and ipd are exact; cost_shown and ipd_shown are how the page shows them. assembled_by_us is 1 when the answer was assembled by us, else 0. A count marked "not published for this model" is not published for that model.
| Argument | Type | Description |
|---|---|---|
| job* | string | Job id (see list_jobs). |
| date | string | YYYY-MM-DD. Default: newest day. |
{
"type": "object",
"$schema": "https://json-schema.org/draft/2020-12/schema",
"required": [
"job"
],
"properties": {
"job": {
"enum": [
"classify",
"convert",
"decide",
"extract",
"find",
"fix",
"match",
"resolve",
"write"
],
"type": "string",
"description": "Job id (see list_jobs)."
},
"date": {
"type": "string",
"description": "YYYY-MM-DD. Default: newest day."
}
}
}Use to follow one model on one job over time (for example, has it got cheaper or worse). Returns its published days: right answers, items, cost per 1,000 right in US dollars and IPD (right answers per dollar). Model names come from list_jobs. On Extract, every row names its test; the answer says what the two tests are, so read cost across 4 Nov with that in mind. Fields: cost_per_1000_right_usd and ipd are exact; cost_shown and ipd_shown are how the page shows them. assembled_by_us is 1 when the answer was assembled by us, else 0. A count marked "not published for this model" is not published for that model.
| Argument | Type | Description |
|---|---|---|
| job* | string | Job id (see list_jobs). |
| model* | string | Model or system name for that job, from list_jobs, e.g. GPT-6 Luna, Claude Haiku 5.5, NL;NL Desk (a system. Not open yet.). |
{
"type": "object",
"$schema": "https://json-schema.org/draft/2020-12/schema",
"required": [
"job",
"model"
],
"properties": {
"job": {
"enum": [
"classify",
"convert",
"decide",
"extract",
"find",
"fix",
"match",
"resolve",
"write"
],
"type": "string",
"description": "Job id (see list_jobs)."
},
"model": {
"type": "string",
"description": "Model or system name for that job, from list_jobs, e.g. GPT-6 Luna, Claude Haiku 5.5, NL;NL Desk (a system. Not open yet.)."
}
}
}Returns the job page's latest published pick. It can be a model or a system. A pick marked "Not open yet." cannot be used yet. Returns the model or system the job page shows as its pick (NL;NL Desk is a system, not a model. Not open yet.), with the page's label and rule, the figures as the page shows them (right answers per $1, cost per 1,000 right), the pick's own date and that model's day row. Fields: cost_per_1000_right_usd and ipd are exact; cost_shown and ipd_shown are how the page shows them. assembled_by_us is 1 when the answer was assembled by us, else 0. A count marked "not published for this model" is not published for that model.
| Argument | Type | Description |
|---|---|---|
| job* | string | Job id (see list_jobs). |
{
"type": "object",
"$schema": "https://json-schema.org/draft/2020-12/schema",
"required": [
"job"
],
"properties": {
"job": {
"enum": [
"classify",
"convert",
"decide",
"extract",
"find",
"fix",
"match",
"resolve",
"write"
],
"type": "string",
"description": "Job id (see list_jobs)."
}
}
}Call this first. Lists the jobs NL;NL Labs tests every day, what each one asks, what counts as right (the site's own words), the models and systems tested on each job (names for the other tools come from here) and the dates covered. NL;NL Labs' own models are marked; NL;NL Desk is a system, not a model. Not open yet.
No arguments.
{
"type": "object",
"properties": {}
}We read every tool's name, description and inputs the way the model will, looking for instructions hidden from you. A clean result means no warning signs, not a guarantee. Scanned Nov 10, 2026.