Skip to content

Commit 8842b6f

Browse files
Das-rebelDas-rebel
committed
docs: copy blog + provider pages to docs/ for GitHub Pages deploy
1 parent 1e07a36 commit 8842b6f

20 files changed

Lines changed: 847 additions & 546 deletions

‎docs/anthropic.html‎

Lines changed: 45 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,45 @@
1+
<!DOCTYPE html>
2+
<html lang="en">
3+
<head>
4+
<meta charset="UTF-8">
5+
<meta name="viewport" content="width=device-width, initial-scale=1.0">
6+
<title>Anthropic Provider — A3M Router</title>
7+
<meta name="description" content="Configure Anthropic Claude models with A3M Router. Claude 3.5 Sonnet, Claude 3 Haiku, rate limits, and configuration.">
8+
<meta name="robots" content="index, follow">
9+
<link rel="canonical" href="https://a3m-router.com/providers/anthropic">
10+
<link rel="stylesheet" href="../assets/styles.css">
11+
</head>
12+
<body>
13+
<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>
14+
<main>
15+
<h1>Anthropic Provider</h1>
16+
<p><strong>Cost Tier:</strong> Paid (~$0.003–$15/1M tokens) · <strong>Rate Limit:</strong> Varies by tier · <strong>MCP Support:</strong> ✅ Native</p>
17+
18+
<h2>Supported Models</h2>
19+
<ul>
20+
<li><strong>Claude 3.5 Sonnet</strong> — Current flagship, best for coding, analysis, and long documents</li>
21+
<li><strong>Claude 3.5 Haiku</strong> — Fast, cost-efficient, for simple tasks</li>
22+
<li><strong>Claude 3 Opus</strong> — Maximum capability, slower and more expensive</li>
23+
<li><strong>Claude 3 Sonnet</strong> — Previous generation, good balance</li>
24+
</ul>
25+
26+
<h2>How A3M Router Routes to Anthropic</h2>
27+
<p>Anthropic models are preferred for:</p>
28+
<ul>
29+
<li>Long-form content generation and summarization</li>
30+
<li>Complex coding and debugging tasks (Claude 3.5 Sonnet excels here)</li>
31+
<li>Safe, helpful assistant interactions</li>
32+
<li>Document analysis with large context windows (200K for Claude 3.5)</li>
33+
</ul>
34+
35+
<h2>Configuration</h2>
36+
<pre><code>ANTHROPIC_API_KEY=sk-ant-...
37+
38+
# Anthropic models are automatically selected when query analysis
39+
# indicates high-quality long-form output or coding tasks</code></pre>
40+
41+
<p><a href="/providers/">← All Providers</a></p>
42+
</main>
43+
<footer><p>&copy; 2025 A3M Router · <a href="https://github.com/Das-rebel/a3m-router">GitHub</a></p></footer>
44+
</body>
45+
</html>

‎docs/best-llm-routers-2025.html‎

Lines changed: 157 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,157 @@
1+
<!DOCTYPE html>
2+
<html lang="en">
3+
<head>
4+
<meta charset="UTF-8">
5+
<meta name="viewport" content="width=device-width, initial-scale=1.0">
6+
<title>Best LLM Routers 2025: A Comprehensive Comparison | A3M Router</title>
7+
<meta name="description" content="Compare the best LLM routers of 2025: A3M Router vs LiteLLM vs Vercel AI Gateway vs LocalAI vs Portkey. Parallel execution, provider count, cost, MCP support, and open source analyzed.">
8+
<meta name="keywords" content="best llm router, llm gateway comparison, litellm vs a3m, vercel ai gateway vs litellm, llm routing 2025, open source llm router">
9+
<meta name="robots" content="index, follow">
10+
<link rel="canonical" href="https://a3m-router.com/blog/best-llm-routers-2025">
11+
<link rel="stylesheet" href="../assets/styles.css">
12+
<script type="application/ld+json">
13+
{
14+
"@context": "https://schema.org",
15+
"@type": "Article",
16+
"headline": "Best LLM Routers 2025: A Comprehensive Comparison",
17+
"description": "A thorough comparison of the leading LLM routing solutions including A3M Router, LiteLLM, Vercel AI Gateway, LocalAI, and Portkey.",
18+
"author": {"@type": "Organization", "name": "A3M Router Team"},
19+
"datePublished": "2025-01-15",
20+
"dateModified": "2025-01-15"
21+
}
22+
</script>
23+
</head>
24+
<body>
25+
<header>
26+
<nav>
27+
<a href="/">A3M Router</a>
28+
<a href="/providers/">Providers</a>
29+
<a href="/blog/">Blog</a>
30+
<a href="https://github.com/Das-rebel/a3m-router">GitHub</a>
31+
</nav>
32+
</header>
33+
34+
<main>
35+
<article>
36+
37+
# Best LLM Routers 2025: A Comprehensive Comparison
38+
39+
*Published January 2025 · 12 min read*
40+
41+
## What is an LLM Router?
42+
43+
An LLM router intelligently directs incoming queries to the most appropriate language model provider based on query characteristics, cost, availability, and capability requirements. Instead of hardcoding a single provider, a router acts as a smart gateway that optimizes for cost, speed, or quality on a per-query basis.
44+
45+
## The Core Difference: Parallel vs Sequential
46+
47+
Most LLM gateways use **sequential fallback**: try Provider A → fail → try Provider B → fail → try Provider C → success → return. The first provider to succeed wins, regardless of cost or quality.
48+
49+
**A3M Router pioneered parallel execution**: call all suitable providers simultaneously → score responses → return the best one. This means you get the optimal answer at the lowest appropriate cost, not just the first one that works.
50+
51+
## Comparison Table
52+
53+
| Router | Parallel Execution | Providers | Open Source | MCP Support | Python SDK | JS/TS SDK | Free Tier |
54+
|--------|:-----------------:|----------:|:-----------:|:-----------:|:----------:|:---------:|:---------:|
55+
| **A3M Router** | ✅ Yes | 47+ | ✅ Apache 2.0 | ✅ | ✅ | ✅ | ✅ |
56+
| LiteLLM | ❌ Sequential | 100+ | ✅ Apache 2.0 | ❌ | ✅ | ✅ | ✅ |
57+
| Vercel AI Gateway | ❌ Sequential | 12+ | ❌ Proprietary | ❌ | ✅ | ✅ | ✅ |
58+
| LocalAI | ❌ Sequential | Local only | ✅ AGPL | ❌ | ✅ | ❌ | ✅ |
59+
| Portkey | ❌ Sequential | 100+ | ❌ Proprietary | ❌ | ✅ | ✅ | ✅ |
60+
| RouteLLM | ❌ Sequential | 6+ | ✅ Apache 2.0 | ❌ | ✅ | ❌ | ✅ |
61+
62+
## Detailed Analysis
63+
64+
### A3M Router
65+
**Best for:** Teams wanting parallel execution with a unique ensemble scoring approach.
66+
67+
A3M Router calls multiple providers simultaneously and uses a weighted scoring system across domain match, specificity, structure alignment, verb matching, and cost tier. The cheapest provider that fully satisfies the query wins. It also features semantic caching, EXP3 diversity routing, and MVT rate-limit rotation.
68+
69+
- **Strengths:** Parallel execution, semantic cache, 47+ providers, open source, MCP server
70+
- **Weaknesses:** Smaller community than LiteLLM
71+
- **Cost:** Free (Apache 2.0); API costs depend on selected providers
72+
73+
### LiteLLM
74+
**Best for:** Maximum provider coverage with simple fallback configuration.
75+
76+
LiteLLM is the most mature open-source option with 100+ provider support. It uses sequential fallback and provides a unified interface for calling any LLM. Very popular (50K+ GitHub stars) with extensive documentation.
77+
78+
- **Strengths:** Largest provider ecosystem, battle-tested, excellent docs
79+
- **Weaknesses:** Sequential fallback only, no parallel execution, no MCP support
80+
- **Cost:** Free to self-host; hosted option available
81+
82+
### Vercel AI Gateway
83+
**Best for:** Teams already on Vercel wanting managed infrastructure.
84+
85+
Vercel's AI Gateway provides routing, caching, and failover for AI applications deployed on Vercel. Simple setup for Vercel users but proprietary and limited to Vercel's infrastructure.
86+
87+
- **Strengths:** Tight Vercel integration, managed service, simple setup
88+
- **Weaknesses:** Proprietary, sequential fallback only, vendor lock-in
89+
- **Cost:** Usage-based via Vercel
90+
91+
### LocalAI
92+
**Best for:** Teams running models entirely on-premise for data privacy.
93+
94+
LocalAI routes to local model instances (Ollama, llama.cpp, etc.) without any external API calls. Fully self-hosted, privacy-first.
95+
96+
- **Strengths:** Complete data privacy, no API costs, fully on-prem
97+
- **Weaknesses:** Requires GPU hardware, sequential only, local models only
98+
- **Cost:** Hardware + electricity
99+
100+
### Portkey
101+
**Best for:** Enterprise teams wanting observability and managed fallback.
102+
103+
Portkey provides a managed gateway with tracking, analytics, and automatic fallback across 100+ providers. Strong observability features.
104+
105+
- **Strengths:** Excellent observability, managed service, 100+ providers
106+
- **Weaknesses:** Sequential fallback, proprietary, enterprise pricing
107+
- **Cost:** Usage-based + subscription
108+
109+
### RouteLLM
110+
**Best for:** Researchers and teams wanting an open-source routing framework.
111+
112+
RouteLLM provides an open-source framework for routing with learning-based approaches. Supports multiple routers and fallback strategies.
113+
114+
- **Strengths:** Open source, research-friendly, multiple routing strategies
115+
- **Weaknesses:** Sequential fallback, smaller provider count, less production-ready
116+
- **Cost:** Free to self-host
117+
118+
## Key Decision Factors
119+
120+
| Use Case | Recommended |
121+
|----------|------------|
122+
| Parallel execution + cost savings | A3M Router |
123+
| Maximum provider variety | LiteLLM |
124+
| Vercel deployment | Vercel AI Gateway |
125+
| Privacy / on-premise | LocalAI |
126+
| Enterprise observability | Portkey |
127+
| Research / experimentation | RouteLLM |
128+
129+
## FAQ
130+
131+
### What is an LLM router?
132+
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.
133+
134+
### How does A3M Router differ from sequential fallback?
135+
A3M Router calls multiple providers in parallel and scores every response using weighted signals (domain match, specificity, structure alignment). The cheapest provider that fully satisfies the query wins — not just the first one to respond successfully.
136+
137+
### Which providers does A3M Router support?
138+
A3M Router supports 47+ providers including OpenAI (GPT-4o, o1), Anthropic (Claude 3.5), Google (Gemini 1.5), Groq (LLaMA 3.3 70B), Mistral, DeepSeek, Cerebras, Ollama, and more.
139+
140+
### Is A3M Router open source?
141+
Yes, A3M Router is Apache 2.0 licensed and available on GitHub at [github.com/Das-rebel/a3m-router](https://github.com/Das-rebel/a3m-router).
142+
143+
### How do I get started with A3M Router?
144+
```bash
145+
npm install -g adaptive-memory-multi-model-router
146+
a3m-router serve
147+
# Then use OpenAI SDK with base_url: http://localhost:8787/v1
148+
```
149+
150+
</article>
151+
</main>
152+
153+
<footer>
154+
<p>&copy; 2025 A3M Router · <a href="https://github.com/Das-rebel/a3m-router">GitHub</a> · <a href="https://www.npmjs.com/package/adaptive-memory-multi-model-router">npm</a></p>
155+
</footer>
156+
</body>
157+
</html>

‎docs/cerebras.html‎

Lines changed: 43 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,43 @@
1+
<!DOCTYPE html>
2+
<html lang="en">
3+
<head>
4+
<meta charset="UTF-8">
5+
<meta name="viewport" content="width=device-width, initial-scale=1.0">
6+
<title>Cerebras Provider — A3M Router</title>
7+
<meta name="description" content="Configure Cerebras (LLaMA 3.3 70B, fastest inference) with A3M Router. Free tier, ultra-fast, and configuration guide.">
8+
<meta name="robots" content="index, follow">
9+
<link rel="canonical" href="https://a3m-router.com/providers/cerebras">
10+
<link rel="stylesheet" href="../assets/styles.css">
11+
</head>
12+
<body>
13+
<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>
14+
<main>
15+
<h1>Cerebras Provider</h1>
16+
<p><strong>Cost Tier:</strong> Free tier available · <strong>Rate Limit:</strong> Generous free tier · <strong>MCP Support:</strong> ✅ Native</p>
17+
18+
<h2>Supported Models</h2>
19+
<ul>
20+
<li><strong>LLaMA 3.3 70B</strong> — Meta's latest, served on Cerebras hardware</li>
21+
<li><strong>LLaMA 3.1 8B</strong> — Fast, lightweight option</li>
22+
<li><strong>Mistral 7B</strong> — Efficient open model</li>
23+
</ul>
24+
25+
<h2>How A3M Router Routes to Cerebras</h2>
26+
<p>Cerebras provides <strong>ultra-fast inference</strong> on large models through its wafer-scale hardware. A3M Router routes here for:</p>
27+
<ul>
28+
<li>Latency-critical simple queries</li>
29+
<li>High-volume free-tier tasks</li>
30+
<li>Situations where Groq's rate limits are hit</li>
31+
</ul>
32+
33+
<h2>Configuration</h2>
34+
<pre><code>CEREBRAS_API_KEY=ccke_...
35+
36+
# Cerebras free tier: generous daily allocation
37+
# Fastest inference available for LLaMA-class models</code></pre>
38+
39+
<p><a href="/providers/">← All Providers</a></p>
40+
</main>
41+
<footer><p>&copy; 2025 A3M Router · <a href="https://github.com/Das-rebel/a3m-router">GitHub</a></p></footer>
42+
</body>
43+
</html>

‎docs/deepseek.html‎

Lines changed: 44 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,44 @@
1+
<!DOCTYPE html>
2+
<html lang="en">
3+
<head>
4+
<meta charset="UTF-8">
5+
<meta name="viewport" content="width=device-width, initial-scale=1.0">
6+
<title>DeepSeek Provider — A3M Router</title>
7+
<meta name="description" content="Configure DeepSeek (V3, DeepSeek Chat) with A3M Router. Chinese AI, low cost, rate limits, and configuration guide.">
8+
<meta name="robots" content="index, follow">
9+
<link rel="canonical" href="https://a3m-router.com/providers/deepseek">
10+
<link rel="stylesheet" href="../assets/styles.css">
11+
</head>
12+
<body>
13+
<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>
14+
<main>
15+
<h1>DeepSeek Provider</h1>
16+
<p><strong>Cost Tier:</strong> Very Low · <strong>Rate Limit:</strong> 60 RPM (free tier) · <strong>MCP Support:</strong> ✅ Native</p>
17+
18+
<h2>Supported Models</h2>
19+
<ul>
20+
<li><strong>DeepSeek V3</strong> — Latest flagship, MoE architecture, 236B parameters</li>
21+
<li><strong>DeepSeek Chat</strong> — General-purpose chat model</li>
22+
<li><strong>DeepSeek Coder</strong> — Specialized for code generation</li>
23+
</ul>
24+
25+
<h2>How A3M Router Routes to DeepSeek</h2>
26+
<p>DeepSeek excels at:</p>
27+
<ul>
28+
<li>Code generation and programming tasks (DeepSeek Coder)</li>
29+
<li>Cost-sensitive production workloads</li>
30+
<li>Multilingual tasks (strong Chinese and English support)</li>
31+
<li>Math and reasoning tasks</li>
32+
</ul>
33+
34+
<h2>Configuration</h2>
35+
<pre><code>DEEPSEEK_API_KEY=sk-dsdk...
36+
37+
# DeepSeek is one of the lowest-cost providers
38+
# A3M Router routes coding and math tasks here for cost savings</code></pre>
39+
40+
<p><a href="/providers/">← All Providers</a></p>
41+
</main>
42+
<footer><p>&copy; 2025 A3M Router · <a href="https://github.com/Das-rebel/a3m-router">GitHub</a></p></footer>
43+
</body>
44+
</html>

‎docs/google.html‎

Lines changed: 47 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,47 @@
1+
<!DOCTYPE html>
2+
<html lang="en">
3+
<head>
4+
<meta charset="UTF-8">
5+
<meta name="viewport" content="width=device-width, initial-scale=1.0">
6+
<title>Google AI Provider — A3M Router</title>
7+
<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.">
8+
<meta name="robots" content="index, follow">
9+
<link rel="canonical" href="https://a3m-router.com/providers/google">
10+
<link rel="stylesheet" href="../assets/styles.css">
11+
</head>
12+
<body>
13+
<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>
14+
<main>
15+
<h1>Google AI Provider</h1>
16+
<p><strong>Cost Tier:</strong> Free tier available · <strong>Rate Limit:</strong> 15–60 RPM depending on tier · <strong>MCP Support:</strong> ✅ Native</p>
17+
18+
<h2>Supported Models</h2>
19+
<ul>
20+
<li><strong>Gemini 1.5 Pro</strong> — Advanced reasoning, 2M token context, multimodal</li>
21+
<li><strong>Gemini 1.5 Flash</strong> — Fast, cost-efficient, 1M context</li>
22+
<li><strong>Gemini 1.0 Pro</strong> — Previous generation</li>
23+
</ul>
24+
25+
<h2>How A3M Router Routes to Google AI</h2>
26+
<p>Google AI is ideal for:</p>
27+
<ul>
28+
<li>Long document processing (up to 2M token context with Gemini 1.5 Pro)</li>
29+
<li>Multimodal inputs (images, video, audio)</li>
30+
<li>Free-tier queries (Gemini Flash has generous free quota)</li>
31+
<li>Fast turnaround on simple Q&amp;A tasks</li>
32+
</ul>
33+
34+
<h2>Configuration</h2>
35+
<pre><code>GOOGLE_API_KEY=AIza...
36+
37+
# Gemini 1.5 Flash has a generous free tier
38+
# A3M Router will route cost-sensitive queries here automatically</code></pre>
39+
40+
<h2>Rate Limits</h2>
41+
<p>Google AI Studio free tier: 15 requests/minute, 1M tokens/month for Gemini 1.5 Flash. Paid tier increases limits significantly.</p>
42+
43+
<p><a href="/providers/">← All Providers</a></p>
44+
</main>
45+
<footer><p>&copy; 2025 A3M Router · <a href="https://github.com/Das-rebel/a3m-router">GitHub</a></p></footer>
46+
</body>
47+
</html>

‎docs/groq.html‎

Lines changed: 44 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,44 @@
1+
<!DOCTYPE html>
2+
<html lang="en">
3+
<head>
4+
<meta charset="UTF-8">
5+
<meta name="viewport" content="width=device-width, initial-scale=1.0">
6+
<title>Groq Provider — A3M Router</title>
7+
<meta name="description" content="Configure Groq (LLaMA 3.3 70B, Mixtral 8x7B) with A3M Router. Free tier, fastest inference, rate limits, and configuration.">
8+
<meta name="robots" content="index, follow">
9+
<link rel="canonical" href="https://a3m-router.com/providers/groq">
10+
<link rel="stylesheet" href="../assets/styles.css">
11+
</head>
12+
<body>
13+
<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>
14+
<main>
15+
<h1>Groq Provider</h1>
16+
<p><strong>Cost Tier:</strong> Free tier available · <strong>Rate Limit:</strong> 14,400 requests/day (free) · <strong>MCP Support:</strong> ✅ Native</p>
17+
18+
<h2>Supported Models</h2>
19+
<ul>
20+
<li><strong>LLaMA 3.3 70B</strong> — Meta's latest open-weight model, excellent quality</li>
21+
<li><strong>Mixtral 8x7B</strong> — Fast mixture-of-experts inference</li>
22+
<li><strong>LLaMA 3.1 8B</strong> — Lightweight, very fast</li>
23+
<li><strong>Gemma 2 9B</strong> — Google's open model</li>
24+
</ul>
25+
26+
<h2>How A3M Router Routes to Groq</h2>
27+
<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>
28+
<ul>
29+
<li>Real-time chat applications where latency is critical</li>
30+
<li>High-volume, simple Q&amp;A (free-tier queries)</li>
31+
<li>Tasks that Groq's models handle well (general reasoning, coding)</li>
32+
</ul>
33+
34+
<h2>Configuration</h2>
35+
<pre><code>GROQ_API_KEY=gsk_...
36+
37+
# Groq free tier: 14,400 requests/day
38+
# A3M Router automatically uses Groq for latency-sensitive queries</code></pre>
39+
40+
<p><a href="/providers/">← All Providers</a></p>
41+
</main>
42+
<footer><p>&copy; 2025 A3M Router · <a href="https://github.com/Das-rebel/a3m-router">GitHub</a></p></footer>
43+
</body>
44+
</html>

0 commit comments

Comments
 (0)