#navbar > div {
  background: #0c3b43 !important;
  color: #fff;
}

#navbar > li > a,
#navbar a {
  color: #fff !important;
}

#navbar > div.z-10.mx-auto.relative > div.hidden.lg\:flex.px-12.h-12 > div > a > div.bg-primary {
  background: #a3ffdc !important;
}

#navbar
  > div.z-10.mx-auto.relative
  > div.relative
  > div.flex.items-center.h-14.py-4.px-5.lg\:hidden
  > div
  > div.font-semibold.text-gray-900.truncate.dark\:text-gray-200 {
  color: #fff !important;
}

#search-bar-entry-mobile > svg {
  background-color: #fff !important;
}

#navbar
  > div.z-10.mx-auto.relative
  > div.relative
  > div.flex.items-center.lg\:px-12.h-16.min-w-0.mx-4.lg\:mx-0
  > div
  > div.flex.lg\:hidden.items-center.gap-2
  > button.h-7.w-5.flex.items-center.justify-end
  > svg {
  background-color: #fff !important;
}

#navbar > div.z-10.mx-auto.relative > div.relative > div.flex.items-center.h-14.py-4.px-5.lg\:hidden > button > svg {
  color: #fff !important;
}

#navbar > li > a div {
  background: #fff !important;
}
.nav-tabs-item {
  color: #fff !important;
}

#topbar-cta-button {
  color: #fff !important;
  background: #0c3b43 !important;
  border: 1px solid #fff !important;
  padding: 1px 10px !important;
  border-radius: 5px !important;
}

#topbar-cta-button span {
  color: #0c3b43 !important;
  font-weight: 600 !important;
}

#topbar-cta-button {
  background: #a3ffdc !important;
  color: #0c3b43 !important;
  border: none !important;
  padding: 2px 16px !important;
  border-radius: 8px !important;
  font-size: 16px !important;
  transition: background-color 0.2s ease !important;
}

#topbar-cta-button:hover {
  background: #66e0b2 !important;
}

#topbar-cta-button svg {
  display: none !important;
}

/* Custom styles for the documentation */
@media (min-width: 1024px) {
  #content-area {
    margin-top: 1.5rem;
  }
}

div.mdx-content {
  margin-top: 4rem;
}

.font-bold {
  font-weight: 600;
}

table thead th {
  text-align: left;
}

p[data-component-part='step-title'] {
  margin-bottom: 0 !important;
}

[data-component-part='step-content'] p:first-child {
  margin-top: 8px !important;
}

:root {
  --font-family-headings-custom: 'Inter', sans-serif !important;
}

header#header .eyebrow {
  color: #3e4040;
  font-weight: 400;
  margin-bottom: 1rem;
}

#header .text-lg {
  font-size: 16px;
}

h5#sidebar-title {
  font-weight: 600;
  font-size: 14px;
  color: #505152;
}

#sidebar-group li > a.font-semibold {
  background-color: #F4F4F5;
  border-radius: 4px;
}

button[type='button'] div.text-gray-900 {
  color: inherit;
}

#table-of-contents-layout > #table-of-contents{
    padding-top: 4rem;
}


<>
<script>
  window.intercomSettings = {
    api_base: "https://api-iam.intercom.io",
    app_id: "glx5mihe",
  };
</script>


<script>
  // We pre-filled your app ID in the widget URL: 'https://widget.intercom.io/widget/glx5mihe'
  (function(){var w=window;var ic=w.Intercom;if(typeof ic==="function"){ic('reattach_activator');ic('update',w.intercomSettings);}else{var d=document;var i=function(){i.c(arguments);};i.q=[];i.c=function(args){i.q.push(args);};w.Intercom=i;var l=function(){var s=d.createElement('script');s.type='text/javascript';s.async=true;s.src='https://widget.intercom.io/widget/glx5mihe';var x=d.getElementsByTagName('script')[0];x.parentNode.insertBefore(s,x);};if(document.readyState==='complete'){l();}else if(w.attachEvent){w.attachEvent('onload',l);}else{w.addEventListener('load',l,false);}}})();
</script>
</>

// Custom script to update page titles with Maxim SDK suffix
(function () {
    function getSDKType() {
        // Check if current page is an SDK reference page and return the SDK type
        const currentPath = window.location.pathname;

        // Check for Python SDK reference pages
        if (currentPath.includes("/sdk/python")) {
            return "python";
        }

        // Check for TypeScript SDK reference pages
        if (currentPath.includes("/sdk/typescript")) {
            return "typescript";
        }

        // Not an SDK page
        return null;
    }

    function updatePageTitle() {
        const sdkType = getSDKType();

        // Only update title if this is an SDK page
        if (!sdkType) {
            return;
        }

        // Get the current page title from the frontmatter or h1
        const titleElement =
            document.querySelector("h1") || document.querySelector("[data-title]");
        let pageTitle = "";

        // Try to get title from various sources
        if (titleElement) {
            pageTitle = titleElement.textContent || titleElement.innerText || "";
        }

        // Fallback to document title if no h1 found
        if (!pageTitle) {
            pageTitle = document.title;
        }

        // Clean up the title (remove extra whitespace)
        pageTitle = pageTitle.trim();

        // Determine the appropriate suffix based on SDK type
        const suffix =
            sdkType === "python" ? "| Maxim Python SDK" : "| Maxim TypeScript SDK";

        // Only update if we have a meaningful title and it doesn't already include our suffix
        if (pageTitle && !pageTitle.includes(suffix)) {
            // Remove any existing SDK suffixes that might be there
            pageTitle = pageTitle.replace(
                /\s*\|\s*Maxim\s+(Python|TypeScript)\s+SDK.*$/,
                "",
            );

            // Also remove any other existing suffixes
            pageTitle = pageTitle.replace(/\s*\|.*$/, "");

            // Add our custom suffix
            document.title = pageTitle + " " + suffix;
        }
    }

    // Update title when page loads
    if (document.readyState === "loading") {
        document.addEventListener("DOMContentLoaded", updatePageTitle);
    } else {
        updatePageTitle();
    }

    // Also update when navigating (for SPAs)
    if (window.history && window.history.pushState) {
        const originalPushState = window.history.pushState;
        window.history.pushState = function () {
            originalPushState.apply(window.history, arguments);
            setTimeout(updatePageTitle, 100); // Small delay to ensure DOM is updated
        };

        const originalReplaceState = window.history.replaceState;
        window.history.replaceState = function () {
            originalReplaceState.apply(window.history, arguments);
            setTimeout(updatePageTitle, 100);
        };

        window.addEventListener("popstate", function () {
            setTimeout(updatePageTitle, 100);
        });
    }

    // Watch for title changes using MutationObserver
    if (typeof MutationObserver !== "undefined") {
        const observer = new MutationObserver(function (mutations) {
            mutations.forEach(function (mutation) {
                if (
                    mutation.type === "childList" ||
                    mutation.type === "characterData"
                ) {
                    // Check if title-related elements changed
                    const titleChanged = Array.from(mutation.addedNodes).some(
                        (node) =>
                            node.nodeType === Node.ELEMENT_NODE &&
                            (node.tagName === "H1" ||
                                (node.querySelector && node.querySelector("h1"))),
                    );

                    if (titleChanged || mutation.target.tagName === "TITLE") {
                        setTimeout(updatePageTitle, 50);
                    }
                }
            });
        });

        observer.observe(document, {
            childList: true,
            subtree: true,
            characterData: true,
        });

        // Also observe the title element specifically
        const titleElement = document.querySelector("title");
        if (titleElement) {
            observer.observe(titleElement, {
                childList: true,
                characterData: true,
            });
        }
    }
})();


names

pass_fail_criteria

evaluate

guarded_evaluate

BaseEvaluator

Base_Evaluator utilities for evaluation tools and utilities for assessing model performance.

Maxim Docs

Maxim streamlines AI application development and deployment by applying traditional software best practices to non-deterministic AI workflows.

Platform Overview

Learn how to get started with your first evaluation run in Maxim

Running Your First Eval

Learn how to evaluate AI application performance through prompt testing, workflow automation, and continuous log monitoring. Streamline your AI testing pipeline with comprehensive evaluation tools.

Offline Evaluation Overview

Offline Evaluation Concepts

Online Evaluation Overview

Learn how to configure notification channels (Slack and PagerDuty) and set up alerts to monitor your AI application's performance and quality metrics.

Set Up Alerts and Notifications

Monitor AI applications in real-time with Maxim's enterprise-grade LLM observability platform.

Tracing Overview

Learn about the key concepts of Maxim's distributed tracing for AI applications.

Tracing Concepts

Set up distributed tracing for your GenAI applications to monitor performance and debug issues across services.

Tracing Quickstart

Learn how to use the dashboard to filter and sort your logs

Dashboard

Learn how to integrate Maxim with external observability platforms using data connectors for enhanced log analysis and monitoring.

OTel Compatibility

Learn how to export your logs and evaluation data in Maxim

Exports

Learn how to set up reporting for your logs and evaluation data in Maxim

Reporting

Simulation Overview

Test your AI's conversational abilities with realistic, scenario-based simulations

Simulation Runs

Library Overview

Explore key concepts in AI evaluation, including evaluators, datasets, and custom tools for assessing model performance and output quality.

Library Concepts

Learn how to create, use, and evaluate context sources for your AI applications

Context Sources

Comprehensive documentation for creating and using different types of prompt tools in Maxim

Prompt Tools

Learn how to create and use prompt partials in Maxim

Creating Prompt Partials

Learn how to create a comparison report for your test runs

Test Runs Comparison Dashboard

Create custom dashboards to analyze and track your AI application logs across repositories using configurable metrics and filters charts

Custom Logs Dashboards

How to integrate Maxim's observability and real-time evaluation capabilities with OpenAI Agents SDK.

OpenAI Agents SDK

Learn how to create a PagerDuty integration in Maxim to receive notifications when your AI application's performance metrics or quality scores exceed specified thresholds.

Create a PagerDuty Integration

Learn how to create a Slack integration in Maxim to receive notifications when your AI application's performance metrics or quality scores exceed specified thresholds.

Create a Slack Integration

Learn how to invite team members and create roles in Maxim.

Members and Roles

Model Configuration

Maxim API keys

Learn how to set up custom token pricing in Maxim for accurate cost reporting in AI evaluations and logs, ensuring displayed costs match your actual expenses.

Custom Pricing

Vault

Learn how to set up two-factor authentication in Maxim.

Two-Factor Authentication

Dive into the Maxim SDK to supercharge your AI application development

Introduction

Overview

Upgrading to v3

Welcome to the Maxim API documentation. This guide provides comprehensive information about our available APIs, their endpoints, and how to use them.

API Reference Overview

Maxim offers self hosting and flexible enterprise deployment options with either full VPC isolation (Zero Touch) or hybrid setup with secure VPC peering (Data Plane), tailored to your security needs.

Self-Hosting Overview

This guide outlines Maxim's zero-touch deployment process, covering infrastructure components, security protocols, and supported cloud providers.

Zero Touch Deployment

This guide details Maxim's data plane deployment process, outlining how to establish data processing infrastructure within your cloud environment. It emphasizes enhanced security, control, and data tenancy, ensuring compliance with data residency requirements while leveraging cloud-based services.

Overview

Python

Typescript

​BaseEvaluator

​names

​pass_fail_criteria

​evaluate

​guarded_evaluate

BaseEvaluator

names

pass_fail_criteria

evaluate

guarded_evaluate