adaptive-memory-multi-model-router 2.16.0 → 2.16.2

package/docs-site/index.html

@@ -6,7 +6,7 @@
 
     <!-- Primary SEO Meta Tags -->
     <title>A3M Router — Parallel LLM Routing Gateway</title>
-    <meta name="description" content="Parallel LLM routing gateway. Routes queries to cheapest capable model across 47+ providers in parallel. 63% cost savings vs premium-only routing. Drop-in OpenAI proxy with 47+ providers.">
+    <meta name="description" content="Intelligent LLM routing proxy. Routes queries to cheapest capable model across 47+ providers. 96.77% RouterArena accuracy, $0.0768 per 1K tokens. Drop-in OpenAI-compatible API.">
     <meta name="keywords" content="llm router benchmark, llm routing accuracy, routellm alternative, litellm alternative, llm cost optimization, openai proxy free, llm gateway open source, lightweight llm router, keyword-based llm routing, drop-in openai proxy, llm routing without gpu, how to reduce openai api costs">
     <meta name="author" content="A3M Router Team">
     <meta name="robots" content="index, follow, max-snippet:-1, max-image-preview:large">
@@ -29,6 +29,64 @@
     <meta name="twitter:description" content="Parallel multi-provider routing with confidence scoring. Semantic cache. 47+ providers. OpenAI-compatible proxy.">
     <meta name="twitter:image" content="https://das-rebel.github.io/adaptive-memory-multi-model-router/assets/og-banner.svg">
 
+    <!-- JSON-LD Structured Data: FAQPage -->
+    <script type="application/ld+json">
+    {
+      "@context": "https://schema.org",
+      "@type": "FAQPage",
+      "mainEntity": [
+        {
+          "@type": "Question",
+          "name": "What is an LLM router?",
+          "acceptedAnswer": {
+            "@type": "Answer",
+            "text": "An LLM router is a gateway that intelligently directs queries to the optimal language model provider based on query characteristics, cost, availability, and capability requirements — rather than hardcoding a single provider. This enables cost optimization, automatic failover, and quality maximization on a per-query basis."
+          }
+        },
+        {
+          "@type": "Question",
+          "name": "How does A3M Router differ from sequential fallback?",
+          "acceptedAnswer": {
+            "@type": "Answer",
+            "text": "Most gateways use sequential fallback: try Provider A, fail, try B, fail, try C, succeed, return. The first provider to succeed wins. A3M Router calls multiple providers in parallel and scores every response using weighted signals (domain match, specificity, structure alignment, verb matching, cost tier). The cheapest provider that fully satisfies the query wins — not just the first one that responds successfully."
+          }
+        },
+        {
+          "@type": "Question",
+          "name": "Which providers does A3M Router support?",
+          "acceptedAnswer": {
+            "@type": "Answer",
+            "text": "A3M Router supports 47+ providers including OpenAI (GPT-4o, GPT-4o-mini, o1-preview, o1-mini), Anthropic (Claude 3.5 Sonnet, Claude 3 Haiku), Google (Gemini 1.5 Pro, Gemini 1.5 Flash), Groq (LLaMA 3.3 70B, Mixtral 8x7B), Mistral (Mistral Large, Mistral 7B), DeepSeek (DeepSeek V3, DeepSeek Chat), Cerebras (LLaMA 3.3 70B), Ollama (local models), and many more."
+          }
+        },
+        {
+          "@type": "Question",
+          "name": "How fast is A3M Router?",
+          "acceptedAnswer": {
+            "@type": "Answer",
+            "text": "A3M Router makes routing decisions in sub-millisecond time (typically 0.1–0.5ms) using a keyword-based classifier. The end-to-end latency depends on the selected provider's model. Parallel calls wait for the fastest responders, so ensemble mode often completes faster than sequential fallback."
+          }
+        },
+        {
+          "@type": "Question",
+          "name": "Is A3M Router open source?",
+          "acceptedAnswer": {
+            "@type": "Answer",
+            "text": "Yes. A3M Router is Apache 2.0 licensed and fully open source. The core routing engine, MCP server, Python SDK, and TypeScript/Node.js packages are all available on GitHub at github.com/Das-rebel/a3m-router."
+          }
+        },
+        {
+          "@type": "Question",
+          "name": "How do I get started with A3M Router?",
+          "acceptedAnswer": {
+            "@type": "Answer",
+            "text": "npm install -g adaptive-memory-multi-model-router && a3m-router serve. Then use the OpenAI SDK with base_url: http://localhost:8787/v1. For Python: pip install adaptive-memory-multi-model-router and use the A3MRouter client. Full docs at https://a3m-router.com."
+          }
+        }
+      ]
+    }
+    </script>
+
     <!-- JSON-LD Structured Data: SoftwareApplication -->
     <script type="application/ld+json">
     {
@@ -468,20 +526,20 @@
             
             <div class="stats">
                 <div class="stat">
-                    <div class="stat-value">2,775</div>
-                    <div class="stat-label">Downloads in 3 Days</div>
+                    <div class="stat-value">5,400+</div>
+                    <div class="stat-label">Monthly Downloads</div>
                 </div>
                 <div class="stat">
-                    <div class="stat-value">245%</div>
-                    <div class="stat-label">Growth, Zero Budget</div>
+                    <div class="stat-value">47+</div>
+                    <div class="stat-label">LLM Providers</div>
                 </div>
                 <div class="stat">
-                    <div class="stat-value">39</div>
-                    <div class="stat-label">LLM Providers</div>
+                    <div class="stat-value">96.77%</div>
+                    <div class="stat-label">RouterArena Accuracy</div>
                 </div>
                 <div class="stat">
-                    <div class="stat-value">67%</div>
-                    <div class="stat-label">Routing Accuracy</div>
+                    <div class="stat-value">$0.08</div>
+                    <div class="stat-label">Per 1K Tokens</div>
                 </div>
             </div>
             
@@ -621,6 +679,7 @@ npx a3m-router serve
             <div class="links">
                 <a href="https://www.npmjs.com/package/adaptive-memory-multi-model-router">NPM</a>
                 <a href="https://github.com/Das-rebel/a3m-router">GitHub</a>
+                <a href="/sitemap.xml">Sitemap</a>
                 <a href="https://github.com/Das-rebel/a3m-router/issues">Issues</a>
                 <a href="https://github.com/Das-rebel/a3m-router/discussions">Discussions</a>
                 <a href="https://github.com/Das-rebel/a3m-router/blob/main/docs/API.md">API Docs</a>

package/docs-site/providers/anthropic.html

@@ -0,0 +1,45 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+    <meta charset="UTF-8">
+    <meta name="viewport" content="width=device-width, initial-scale=1.0">
+    <title>Anthropic Provider — A3M Router</title>
+    <meta name="description" content="Configure Anthropic Claude models with A3M Router. Claude 3.5 Sonnet, Claude 3 Haiku, rate limits, and configuration.">
+    <meta name="robots" content="index, follow">
+    <link rel="canonical" href="https://a3m-router.com/providers/anthropic">
+    <link rel="stylesheet" href="../assets/styles.css">
+</head>
+<body>
+<header><nav><a href="/">A3M Router</a> · <a href="/providers/">All Providers</a> · <a href="https://github.com/Das-rebel/a3m-router">GitHub</a></nav></header>
+<main>
+<h1>Anthropic Provider</h1>
+<p><strong>Cost Tier:</strong> Paid (~$0.003–$15/1M tokens) · <strong>Rate Limit:</strong> Varies by tier · <strong>MCP Support:</strong> ✅ Native</p>
+
+<h2>Supported Models</h2>
+<ul>
+<li><strong>Claude 3.5 Sonnet</strong> — Current flagship, best for coding, analysis, and long documents</li>
+<li><strong>Claude 3.5 Haiku</strong> — Fast, cost-efficient, for simple tasks</li>
+<li><strong>Claude 3 Opus</strong> — Maximum capability, slower and more expensive</li>
+<li><strong>Claude 3 Sonnet</strong> — Previous generation, good balance</li>
+</ul>
+
+<h2>How A3M Router Routes to Anthropic</h2>
+<p>Anthropic models are preferred for:</p>
+<ul>
+<li>Long-form content generation and summarization</li>
+<li>Complex coding and debugging tasks (Claude 3.5 Sonnet excels here)</li>
+<li>Safe, helpful assistant interactions</li>
+<li>Document analysis with large context windows (200K for Claude 3.5)</li>
+</ul>
+
+<h2>Configuration</h2>
+<pre><code>ANTHROPIC_API_KEY=sk-ant-...
+
+# Anthropic models are automatically selected when query analysis
+# indicates high-quality long-form output or coding tasks</code></pre>
+
+<p><a href="/providers/">← All Providers</a></p>
+</main>
+<footer><p>&copy; 2025 A3M Router · <a href="https://github.com/Das-rebel/a3m-router">GitHub</a></p></footer>
+</body>
+</html>

package/docs-site/providers/cerebras.html

@@ -0,0 +1,43 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+    <meta charset="UTF-8">
+    <meta name="viewport" content="width=device-width, initial-scale=1.0">
+    <title>Cerebras Provider — A3M Router</title>
+    <meta name="description" content="Configure Cerebras (LLaMA 3.3 70B, fastest inference) with A3M Router. Free tier, ultra-fast, and configuration guide.">
+    <meta name="robots" content="index, follow">
+    <link rel="canonical" href="https://a3m-router.com/providers/cerebras">
+    <link rel="stylesheet" href="../assets/styles.css">
+</head>
+<body>
+<header><nav><a href="/">A3M Router</a> · <a href="/providers/">All Providers</a> · <a href="https://github.com/Das-rebel/a3m-router">GitHub</a></nav></header>
+<main>
+<h1>Cerebras Provider</h1>
+<p><strong>Cost Tier:</strong> Free tier available · <strong>Rate Limit:</strong> Generous free tier · <strong>MCP Support:</strong> ✅ Native</p>
+
+<h2>Supported Models</h2>
+<ul>
+<li><strong>LLaMA 3.3 70B</strong> — Meta's latest, served on Cerebras hardware</li>
+<li><strong>LLaMA 3.1 8B</strong> — Fast, lightweight option</li>
+<li><strong>Mistral 7B</strong> — Efficient open model</li>
+</ul>
+
+<h2>How A3M Router Routes to Cerebras</h2>
+<p>Cerebras provides <strong>ultra-fast inference</strong> on large models through its wafer-scale hardware. A3M Router routes here for:</p>
+<ul>
+<li>Latency-critical simple queries</li>
+<li>High-volume free-tier tasks</li>
+<li>Situations where Groq's rate limits are hit</li>
+</ul>
+
+<h2>Configuration</h2>
+<pre><code>CEREBRAS_API_KEY=ccke_...
+
+# Cerebras free tier: generous daily allocation
+# Fastest inference available for LLaMA-class models</code></pre>
+
+<p><a href="/providers/">← All Providers</a></p>
+</main>
+<footer><p>&copy; 2025 A3M Router · <a href="https://github.com/Das-rebel/a3m-router">GitHub</a></p></footer>
+</body>
+</html>

package/docs-site/providers/deepseek.html

@@ -0,0 +1,44 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+    <meta charset="UTF-8">
+    <meta name="viewport" content="width=device-width, initial-scale=1.0">
+    <title>DeepSeek Provider — A3M Router</title>
+    <meta name="description" content="Configure DeepSeek (V3, DeepSeek Chat) with A3M Router. Chinese AI, low cost, rate limits, and configuration guide.">
+    <meta name="robots" content="index, follow">
+    <link rel="canonical" href="https://a3m-router.com/providers/deepseek">
+    <link rel="stylesheet" href="../assets/styles.css">
+</head>
+<body>
+<header><nav><a href="/">A3M Router</a> · <a href="/providers/">All Providers</a> · <a href="https://github.com/Das-rebel/a3m-router">GitHub</a></nav></header>
+<main>
+<h1>DeepSeek Provider</h1>
+<p><strong>Cost Tier:</strong> Very Low · <strong>Rate Limit:</strong> 60 RPM (free tier) · <strong>MCP Support:</strong> ✅ Native</p>
+
+<h2>Supported Models</h2>
+<ul>
+<li><strong>DeepSeek V3</strong> — Latest flagship, MoE architecture, 236B parameters</li>
+<li><strong>DeepSeek Chat</strong> — General-purpose chat model</li>
+<li><strong>DeepSeek Coder</strong> — Specialized for code generation</li>
+</ul>
+
+<h2>How A3M Router Routes to DeepSeek</h2>
+<p>DeepSeek excels at:</p>
+<ul>
+<li>Code generation and programming tasks (DeepSeek Coder)</li>
+<li>Cost-sensitive production workloads</li>
+<li>Multilingual tasks (strong Chinese and English support)</li>
+<li>Math and reasoning tasks</li>
+</ul>
+
+<h2>Configuration</h2>
+<pre><code>DEEPSEEK_API_KEY=sk-dsdk...
+
+# DeepSeek is one of the lowest-cost providers
+# A3M Router routes coding and math tasks here for cost savings</code></pre>
+
+<p><a href="/providers/">← All Providers</a></p>
+</main>
+<footer><p>&copy; 2025 A3M Router · <a href="https://github.com/Das-rebel/a3m-router">GitHub</a></p></footer>
+</body>
+</html>

package/docs-site/providers/google.html

@@ -0,0 +1,47 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+    <meta charset="UTF-8">
+    <meta name="viewport" content="width=device-width, initial-scale=1.0">
+    <title>Google AI Provider — A3M Router</title>
+    <meta name="description" content="Configure Google AI (Gemini 1.5 Pro, Gemini 1.5 Flash) with A3M Router. Free tier, rate limits, and configuration guide.">
+    <meta name="robots" content="index, follow">
+    <link rel="canonical" href="https://a3m-router.com/providers/google">
+    <link rel="stylesheet" href="../assets/styles.css">
+</head>
+<body>
+<header><nav><a href="/">A3M Router</a> · <a href="/providers/">All Providers</a> · <a href="https://github.com/Das-rebel/a3m-router">GitHub</a></nav></header>
+<main>
+<h1>Google AI Provider</h1>
+<p><strong>Cost Tier:</strong> Free tier available · <strong>Rate Limit:</strong> 15–60 RPM depending on tier · <strong>MCP Support:</strong> ✅ Native</p>
+
+<h2>Supported Models</h2>
+<ul>
+<li><strong>Gemini 1.5 Pro</strong> — Advanced reasoning, 2M token context, multimodal</li>
+<li><strong>Gemini 1.5 Flash</strong> — Fast, cost-efficient, 1M context</li>
+<li><strong>Gemini 1.0 Pro</strong> — Previous generation</li>
+</ul>
+
+<h2>How A3M Router Routes to Google AI</h2>
+<p>Google AI is ideal for:</p>
+<ul>
+<li>Long document processing (up to 2M token context with Gemini 1.5 Pro)</li>
+<li>Multimodal inputs (images, video, audio)</li>
+<li>Free-tier queries (Gemini Flash has generous free quota)</li>
+<li>Fast turnaround on simple Q&amp;A tasks</li>
+</ul>
+
+<h2>Configuration</h2>
+<pre><code>GOOGLE_API_KEY=AIza...
+
+# Gemini 1.5 Flash has a generous free tier
+# A3M Router will route cost-sensitive queries here automatically</code></pre>
+
+<h2>Rate Limits</h2>
+<p>Google AI Studio free tier: 15 requests/minute, 1M tokens/month for Gemini 1.5 Flash. Paid tier increases limits significantly.</p>
+
+<p><a href="/providers/">← All Providers</a></p>
+</main>
+<footer><p>&copy; 2025 A3M Router · <a href="https://github.com/Das-rebel/a3m-router">GitHub</a></p></footer>
+</body>
+</html>

package/docs-site/providers/groq.html

@@ -0,0 +1,44 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+    <meta charset="UTF-8">
+    <meta name="viewport" content="width=device-width, initial-scale=1.0">
+    <title>Groq Provider — A3M Router</title>
+    <meta name="description" content="Configure Groq (LLaMA 3.3 70B, Mixtral 8x7B) with A3M Router. Free tier, fastest inference, rate limits, and configuration.">
+    <meta name="robots" content="index, follow">
+    <link rel="canonical" href="https://a3m-router.com/providers/groq">
+    <link rel="stylesheet" href="../assets/styles.css">
+</head>
+<body>
+<header><nav><a href="/">A3M Router</a> · <a href="/providers/">All Providers</a> · <a href="https://github.com/Das-rebel/a3m-router">GitHub</a></nav></header>
+<main>
+<h1>Groq Provider</h1>
+<p><strong>Cost Tier:</strong> Free tier available · <strong>Rate Limit:</strong> 14,400 requests/day (free) · <strong>MCP Support:</strong> ✅ Native</p>
+
+<h2>Supported Models</h2>
+<ul>
+<li><strong>LLaMA 3.3 70B</strong> — Meta's latest open-weight model, excellent quality</li>
+<li><strong>Mixtral 8x7B</strong> — Fast mixture-of-experts inference</li>
+<li><strong>LLaMA 3.1 8B</strong> — Lightweight, very fast</li>
+<li><strong>Gemma 2 9B</strong> — Google's open model</li>
+</ul>
+
+<h2>How A3M Router Routes to Groq</h2>
+<p>Groq's key advantage is <strong>inference speed</strong> — the fastest in the industry for supported models. A3M Router routes to Groq for:</p>
+<ul>
+<li>Real-time chat applications where latency is critical</li>
+<li>High-volume, simple Q&amp;A (free-tier queries)</li>
+<li>Tasks that Groq's models handle well (general reasoning, coding)</li>
+</ul>
+
+<h2>Configuration</h2>
+<pre><code>GROQ_API_KEY=gsk_...
+
+# Groq free tier: 14,400 requests/day
+# A3M Router automatically uses Groq for latency-sensitive queries</code></pre>
+
+<p><a href="/providers/">← All Providers</a></p>
+</main>
+<footer><p>&copy; 2025 A3M Router · <a href="https://github.com/Das-rebel/a3m-router">GitHub</a></p></footer>
+</body>
+</html>

package/docs-site/providers/index.html

@@ -0,0 +1,41 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+    <meta charset="UTF-8">
+    <meta name="viewport" content="width=device-width, initial-scale=1.0">
+    <title>Providers — A3M Router</title>
+    <meta name="description" content="All LLM providers supported by A3M Router: OpenAI, Anthropic, Google, Groq, Mistral, Ollama, DeepSeek, Cerebras, and more.">
+    <meta name="robots" content="index, follow">
+    <link rel="canonical" href="https://a3m-router.com/providers/">
+    <link rel="stylesheet" href="../assets/styles.css">
+</head>
+<body>
+<header><nav><a href="/">A3M Router</a> · <a href="/blog/">Blog</a> · <a href="https://github.com/Das-rebel/a3m-router">GitHub</a></nav></header>
+<main>
+<h1>Supported Providers</h1>
+<p>A3M Router supports 47+ providers. Here are the most commonly used:</p>
+
+<h2>Free Tier Providers</h2>
+<ul>
+<li><a href="google.html">Google AI</a> — Gemini 1.5 Flash, generous free tier</li>
+<li><a href="groq.html">Groq</a> — LLaMA 3.3 70B, fastest inference, free tier</li>
+<li><a href="cerebras.html">Cerebras</a> — LLaMA 3.3 70B, ultra-fast, free tier</li>
+<li><a href="ollama.html">Ollama</a> — Local models, zero API cost</li>
+</ul>
+
+<h2>Premium Providers</h2>
+<ul>
+<li><a href="openai.html">OpenAI</a> — GPT-4o, o1, GPT-4o-mini</li>
+<li><a href="anthropic.html">Anthropic</a> — Claude 3.5 Sonnet, Claude 3 Haiku</li>
+<li><a href="mistral.html">Mistral AI</a> — Mistral Large, Codestral</li>
+<li><a href="deepseek.html">DeepSeek</a> — DeepSeek V3, DeepSeek Coder</li>
+</ul>
+
+<h2>Adding Your Provider</h2>
+<p>A3M Router is designed to work with any OpenAI-compatible API. To add a new provider, set the appropriate environment variable and A3M Router will automatically detect and use it.</p>
+
+<p><a href="/">← Back to Home</a></p>
+</main>
+<footer><p>&copy; 2025 A3M Router · <a href="https://github.com/Das-rebel/a3m-router">GitHub</a></p></footer>
+</body>
+</html>

package/docs-site/providers/mistral.html

@@ -0,0 +1,43 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+    <meta charset="UTF-8">
+    <meta name="viewport" content="width=device-width, initial-scale=1.0">
+    <title>Mistral AI Provider — A3M Router</title>
+    <meta name="description" content="Configure Mistral AI (Mistral Large, Mistral 7B) with A3M Router. European AI, rate limits, and configuration.">
+    <meta name="robots" content="index, follow">
+    <link rel="canonical" href="https://a3m-router.com/providers/mistral">
+    <link rel="stylesheet" href="../assets/styles.css">
+</head>
+<body>
+<header><nav><a href="/">A3M Router</a> · <a href="/providers/">All Providers</a> · <a href="https://github.com/Das-rebel/a3m-router">GitHub</a></nav></header>
+<main>
+<h1>Mistral AI Provider</h1>
+<p><strong>Cost Tier:</strong> Paid · <strong>Rate Limit:</strong> Varies by tier · <strong>MCP Support:</strong> ✅ Native</p>
+
+<h2>Supported Models</h2>
+<ul>
+<li><strong>Mistral Large</strong> — Flagship model for complex reasoning, coding, and analysis</li>
+<li><strong>Mistral 7B</strong> — Fast, efficient open model</li>
+<li><strong>Mixtral 8x22B</strong> — Mixture of experts, excellent quality</li>
+<li><strong>Codestral</strong> — Specialized for code generation</li>
+</ul>
+
+<h2>How A3M Router Routes to Mistral</h2>
+<p>Mistral is a strong choice for:</p>
+<ul>
+<li>European data residency requirements (Mistral is based in France)</li>
+<li>Code generation tasks (Codestral)</li>
+<li>Balanced quality/cost for general reasoning</li>
+</ul>
+
+<h2>Configuration</h2>
+<pre><code>MISTRAL_API_KEY=...
+
+# Set in environment or .env file</code></pre>
+
+<p><a href="/providers/">← All Providers</a></p>
+</main>
+<footer><p>&copy; 2025 A3M Router · <a href="https://github.com/Das-rebel/a3m-router">GitHub</a></p></footer>
+</body>
+</html>

package/docs-site/providers/ollama.html

@@ -0,0 +1,50 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+    <meta charset="UTF-8">
+    <meta name="viewport" content="width=device-width, initial-scale=1.0">
+    <title>Ollama Provider — A3M Router</title>
+    <meta name="description" content="Configure Ollama (local LLM inference) with A3M Router. Zero API costs, privacy-first, and configuration guide.">
+    <meta name="robots" content="index, follow">
+    <link rel="canonical" href="https://a3m-router.com/providers/ollama">
+    <link rel="stylesheet" href="../assets/styles.css">
+</head>
+<body>
+<header><nav><a href="/">A3M Router</a> · <a href="/providers/">All Providers</a> · <a href="https://github.com/Das-rebel/a3m-router">GitHub</a></nav></header>
+<main>
+<h1>Ollama Provider</h1>
+<p><strong>Cost Tier:</strong> Free (hardware only) · <strong>Rate Limit:</strong> GPU-dependent · <strong>MCP Support:</strong> ✅ Native</p>
+
+<h2>Supported Models</h2>
+<ul>
+<li><strong>LLaMA 3.1</strong> — Meta's open instruction-tuned model (8B, 70B)</li>
+<li><strong>Mistral</strong> — Efficient open model</li>
+<li><strong>Qwen 2.5</strong> — Alibaba's multilingual model</li>
+<li><strong>Phi-3</strong> — Microsoft's efficient small model</li>
+<li><strong>Any GGUF-format model</strong> — Custom model support</li>
+</ul>
+
+<h2>How A3M Router Routes to Ollama</h2>
+<p>Ollama provides completely free, local inference. A3M Router routes to Ollama for:</p>
+<ul>
+<li>Maximum privacy (no data leaves your infrastructure)</li>
+<li>Zero API costs for high-volume simple tasks</li>
+<li>Development and testing without API key requirements</li>
+</ul>
+
+<h2>Configuration</h2>
+<pre><code># Install Ollama, then pull models:
+# ollama pull llama3.1
+# ollama pull mistral
+
+# A3M Router connects to local Ollama at http://localhost:11434
+# No API key required — zero cost for inference</code></pre>
+
+<h2>Requirements</h2>
+<p>GPU recommended for reasonable inference speed. CPU-only works for small models (7B params) but is significantly slower.</p>
+
+<p><a href="/providers/">← All Providers</a></p>
+</main>
+<footer><p>&copy; 2025 A3M Router · <a href="https://github.com/Das-rebel/a3m-router">GitHub</a></p></footer>
+</body>
+</html>

package/docs-site/providers/openai.html

@@ -0,0 +1,57 @@
+<!DOCTYPE html>
+<html lang="en">
+<head>
+    <meta charset="UTF-8">
+    <meta name="viewport" content="width=device-width, initial-scale=1.0">
+    <title>OpenAI Provider — A3M Router</title>
+    <meta name="description" content="Configure OpenAI (GPT-4o, o1, GPT-4o-mini) with A3M Router. Cost tier, rate limits, models, and configuration guide.">
+    <meta name="robots" content="index, follow">
+    <link rel="canonical" href="https://a3m-router.com/providers/openai">
+    <link rel="stylesheet" href="../assets/styles.css">
+</head>
+<body>
+<header><nav><a href="/">A3M Router</a> · <a href="/providers/">All Providers</a> · <a href="https://github.com/Das-rebel/a3m-router">GitHub</a></nav></header>
+<main>
+<h1>OpenAI Provider</h1>
+<p><strong>Cost Tier:</strong> Paid (~$0.002–$15/1M tokens) · <strong>Rate Limit:</strong> Varies by model and tier · <strong>MCP Support:</strong> ✅ Native</p>
+
+<h2>Supported Models</h2>
+<ul>
+<li><strong>GPT-4o</strong> — Latest flagship, multimodal, 128K context</li>
+<li><strong>GPT-4o-mini</strong> — Cost-optimized, fast, same capabilities</li>
+<li><strong>o1-preview</strong> — Reasoning model for complex tasks</li>
+<li><strong>o1-mini</strong> — Reasoning model, lower cost</li>
+<li><strong>GPT-4 Turbo</strong> — Previous flagship, 128K context</li>
+</ul>
+
+<h2>How A3M Router Routes to OpenAI</h2>
+<p>A3M Router automatically detects queries requiring reasoning, coding, or multimodal capabilities and routes them to OpenAI models when the query demands it. OpenAI is typically used for:</p>
+<ul>
+<li>Complex reasoning and chain-of-thought tasks (o1 series)</li>
+<li>Code generation and debugging (GPT-4o)</li>
+<li>Multimodal inputs (images + text)</li>
+<li>High-precision instruction following</li>
+</ul>
+
+<h2>Configuration</h2>
+<pre><code># Set in your environment or .env file
+OPENAI_API_KEY=sk-...
+
+# A3M Router automatically detects and uses OpenAI when appropriate
+# No additional config needed for standard use</code></pre>
+
+<h2>Rate Limits</h2>
+<p>Rate limits vary by OpenAI API tier. A3M Router's MVT (Maximum Value-Targeted) rotation automatically distributes requests across multiple API keys if configured, reducing rate limit impact.</p>
+
+<h2>Cost Optimization Tips</h2>
+<ul>
+<li>Use <code>gpt-4o-mini</code> for simple queries — 60x cheaper than GPT-4o</li>
+<li>Enable semantic caching to avoid repeat API calls</li>
+<li>Route straightforward Q&amp;A to free-tier providers (Groq, Cerebras) and reserve OpenAI for complex tasks</li>
+</ul>
+
+<p><a href="/providers/">← All Providers</a></p>
+</main>
+<footer><p>&copy; 2025 A3M Router · <a href="https://github.com/Das-rebel/a3m-router">GitHub</a></p></footer>
+</body>
+</html>

package/docs-site/sitemap.xml

@@ -0,0 +1,69 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
+  <url>
+    <loc>https://a3m-router.com/</loc>
+    <priority>1.0</priority>
+    <changefreq>weekly</changefreq>
+    <lastmod>2025-01-15</lastmod>
+  </url>
+  <url>
+    <loc>https://a3m-router.com/blog/best-llm-routers-2025</loc>
+    <priority>0.8</priority>
+    <changefreq>monthly</changefreq>
+    <lastmod>2025-01-15</lastmod>
+  </url>
+  <url>
+    <loc>https://a3m-router.com/providers/</loc>
+    <priority>0.7</priority>
+    <changefreq>weekly</changefreq>
+    <lastmod>2025-01-15</lastmod>
+  </url>
+  <url>
+    <loc>https://a3m-router.com/providers/openai/</loc>
+    <priority>0.6</priority>
+    <changefreq>monthly</changefreq>
+    <lastmod>2025-01-15</lastmod>
+  </url>
+  <url>
+    <loc>https://a3m-router.com/providers/anthropic/</loc>
+    <priority>0.6</priority>
+    <changefreq>monthly</changefreq>
+    <lastmod>2025-01-15</lastmod>
+  </url>
+  <url>
+    <loc>https://a3m-router.com/providers/google/</loc>
+    <priority>0.6</priority>
+    <changefreq>monthly</changefreq>
+    <lastmod>2025-01-15</lastmod>
+  </url>
+  <url>
+    <loc>https://a3m-router.com/providers/groq/</loc>
+    <priority>0.6</priority>
+    <changefreq>monthly</changefreq>
+    <lastmod>2025-01-15</lastmod>
+  </url>
+  <url>
+    <loc>https://a3m-router.com/providers/mistral/</loc>
+    <priority>0.6</priority>
+    <changefreq>monthly</changefreq>
+    <lastmod>2025-01-15</lastmod>
+  </url>
+  <url>
+    <loc>https://a3m-router.com/providers/ollama/</loc>
+    <priority>0.6</priority>
+    <changefreq>monthly</changefreq>
+    <lastmod>2025-01-15</lastmod>
+  </url>
+  <url>
+    <loc>https://a3m-router.com/providers/deepseek/</loc>
+    <priority>0.6</priority>
+    <changefreq>monthly</changefreq>
+    <lastmod>2025-01-15</lastmod>
+  </url>
+  <url>
+    <loc>https://a3m-router.com/providers/cerebras/</loc>
+    <priority>0.6</priority>
+    <changefreq>monthly</changefreq>
+    <lastmod>2025-01-15</lastmod>
+  </url>
+</urlset>