adaptive-memory-multi-model-router 2.14.56 β†’ 2.14.57

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (10) hide show
  1. package/README.md +31 -0
  2. package/assets/banner.svg +9 -9
  3. package/assets/social-preview.svg +8 -8
  4. package/docs/QUICK_START_VISIBILITY.md +1 -1
  5. package/docs/SEO_AUDIT.md +21 -23
  6. package/docs/geo/GENERATIVE_ENGINE_OPTIMIZATION.md +4 -3
  7. package/docs-site/assets/og-banner.svg +34 -35
  8. package/docs-site/index.html +24 -24
  9. package/index.html +2 -2
  10. package/package.json +1 -1
package/README.md CHANGED
@@ -1,3 +1,34 @@
1
+ ## ⚑ 30-second install
2
+
3
+ A3M Router is an OpenAI-compatible LLM gateway. Install it, start the proxy, and point your existing OpenAI SDK to `http://localhost:8787/v1`.
4
+
5
+ ```bash
6
+ npm install adaptive-memory-multi-model-router
7
+ npx a3m-router serve
8
+ ```
9
+
10
+ ```python
11
+ from openai import OpenAI
12
+
13
+ client = OpenAI(base_url="http://localhost:8787/v1", api_key="not-needed")
14
+
15
+ response = client.chat.completions.create(
16
+ model="auto", # A3M routes to the cheapest capable provider
17
+ messages=[{"role": "user", "content": "Explain quantum computing in 3 bullets"}]
18
+ )
19
+
20
+ print(response.choices[0].message.content)
21
+ ```
22
+
23
+ In 30 seconds you get:
24
+
25
+ - OpenAI-compatible proxy at `http://localhost:8787/v1`
26
+ - Auto-routing across **47+ providers**
27
+ - Semantic cache, provider health, budget controls, and circuit breakers
28
+ - RouterArena PR #144 proof: **0.9404 score**, **96.77% accuracy**, **$0.0768/1K**, **1.0000 robustness**, **0 abnormal entries** across **8,400 queries**
29
+
30
+ No ML training. No GPU. Drop-in for existing LLM apps.
31
+
1
32
  [πŸ‡¨πŸ‡³ δΈ­ζ–‡](./README_zh.md) Β· [πŸ‡―πŸ‡΅ ζ—₯本θͺž](./README_ja.md) Β· [English](./README.md)
2
33
 
3
34
  ## πŸ†• What's New (v2.14 β€” June 2026)
package/assets/banner.svg CHANGED
@@ -72,36 +72,36 @@
72
72
  <text x="600" y="210" font-family="system-ui, -apple-system, sans-serif" font-size="28" fill="url(#accentGradient)" text-anchor="middle">Adaptive Memory Multi-Model Router</text>
73
73
 
74
74
  <!-- Tagline -->
75
- <text x="600" y="250" font-family="system-ui, -apple-system, sans-serif" font-size="18" fill="#94a3b8" text-anchor="middle">Route LLM queries to optimal providers β€’ Save 50-80% on API costs</text>
75
+ <text x="600" y="250" font-family="system-ui, -apple-system, sans-serif" font-size="18" fill="#94a3b8" text-anchor="middle">No. 1 RouterArena accuracy, cost & robustness among known public baselines</text>
76
76
 
77
77
  <!-- Stats Bar -->
78
78
  <g transform="translate(600, 320)">
79
79
  <rect x="-350" y="0" width="700" height="50" rx="25" fill="rgba(255,255,255,0.05)" stroke="rgba(255,255,255,0.1)" stroke-width="1"/>
80
80
 
81
81
  <!-- Stat 1 -->
82
- <text x="-280" y="32" font-family="system-ui, -apple-system, sans-serif" font-size="14" font-weight="bold" fill="#6366f1" text-anchor="middle">872+</text>
83
- <text x="-280" y="48" font-family="system-ui, -apple-system, sans-serif" font-size="10" fill="#64748b" text-anchor="middle">Weekly Downloads</text>
82
+ <text x="-280" y="32" font-family="system-ui, -apple-system, sans-serif" font-size="14" font-weight="bold" fill="#6366f1" text-anchor="middle">0.9404</text>
83
+ <text x="-280" y="48" font-family="system-ui, -apple-system, sans-serif" font-size="10" fill="#64748b" text-anchor="middle">RouterArena Score</text>
84
84
 
85
85
  <!-- Divider -->
86
86
  <line x1="-200" y1="10" x2="-200" y2="40" stroke="rgba(255,255,255,0.1)" stroke-width="1"/>
87
87
 
88
88
  <!-- Stat 2 -->
89
- <text x="-100" y="32" font-family="system-ui, -apple-system, sans-serif" font-size="14" font-weight="bold" fill="#10b981" text-anchor="middle">12</text>
90
- <text x="-100" y="48" font-family="system-ui, -apple-system, sans-serif" font-size="10" fill="#64748b" text-anchor="middle">LLM Providers</text>
89
+ <text x="-100" y="32" font-family="system-ui, -apple-system, sans-serif" font-size="14" font-weight="bold" fill="#10b981" text-anchor="middle">96.77%</text>
90
+ <text x="-100" y="48" font-family="system-ui, -apple-system, sans-serif" font-size="10" fill="#64748b" text-anchor="middle">Accuracy</text>
91
91
 
92
92
  <!-- Divider -->
93
93
  <line x1="0" y1="10" x2="0" y2="40" stroke="rgba(255,255,255,0.1)" stroke-width="1"/>
94
94
 
95
95
  <!-- Stat 3 -->
96
- <text x="100" y="32" font-family="system-ui, -apple-system, sans-serif" font-size="14" font-weight="bold" fill="#f59e0b" text-anchor="middle">116</text>
97
- <text x="100" y="48" font-family="system-ui, -apple-system, sans-serif" font-size="10" fill="#64748b" text-anchor="middle">Integrations</text>
96
+ <text x="100" y="32" font-family="system-ui, -apple-system, sans-serif" font-size="14" font-weight="bold" fill="#f59e0b" text-anchor="middle">$0.0768/1K</text>
97
+ <text x="100" y="48" font-family="system-ui, -apple-system, sans-serif" font-size="10" fill="#64748b" text-anchor="middle">Cost</text>
98
98
 
99
99
  <!-- Divider -->
100
100
  <line x1="200" y1="10" x2="200" y2="40" stroke="rgba(255,255,255,0.1)" stroke-width="1"/>
101
101
 
102
102
  <!-- Stat 4 -->
103
- <text x="280" y="32" font-family="system-ui, -apple-system, sans-serif" font-size="14" font-weight="bold" fill="#8b5cf6" text-anchor="middle">33</text>
104
- <text x="280" y="48" font-family="system-ui, -apple-system, sans-serif" font-size="10" fill="#64748b" text-anchor="middle">Tests Passing</text>
103
+ <text x="280" y="32" font-family="system-ui, -apple-system, sans-serif" font-size="14" font-weight="bold" fill="#8b5cf6" text-anchor="middle">1.0000</text>
104
+ <text x="280" y="48" font-family="system-ui, -apple-system, sans-serif" font-size="10" fill="#64748b" text-anchor="middle">Robustness</text>
105
105
  </g>
106
106
 
107
107
  <!-- Bottom Gradient Line -->
package/assets/social-preview.svg CHANGED
@@ -79,7 +79,7 @@
79
79
 
80
80
  <!-- Subtitle -->
81
81
  <text x="120" y="170" font-family="'SF Pro Text', -apple-system, 'Segoe UI', Helvetica, Arial, sans-serif" font-size="24" fill="#8b949e" font-weight="400">
82
- Drop-in OpenAI proxy Β· 39 providers Β· 245% growth in 3 days
82
+ Drop-in OpenAI proxy Β· 47+ providers Β· RouterArena PR #144
83
83
  </text>
84
84
 
85
85
  <!-- Stat cards row -->
@@ -88,13 +88,13 @@
88
88
  <rect x="120" y="210" width="320" height="160" rx="16" fill="#161b22" stroke="#238636" stroke-width="2" opacity="0.95"/>
89
89
  <rect x="120" y="210" width="320" height="4" rx="2" fill="url(#greenGrad)"/>
90
90
  <text x="280" y="268" font-family="'SF Mono', 'Fira Code', 'Consolas', monospace" font-size="52" font-weight="800" fill="#2ea043" text-anchor="middle">
91
- 2,775
91
+ 0.9404
92
92
  </text>
93
93
  <text x="280" y="305" font-family="'SF Pro Text', -apple-system, 'Segoe UI', Helvetica, Arial, sans-serif" font-size="16" fill="#8b949e" text-anchor="middle" font-weight="500">
94
- npm downloads
94
+ RouterArena score
95
95
  </text>
96
96
  <text x="280" y="340" font-family="'SF Pro Text', -apple-system, 'Segoe UI', Helvetica, Arial, sans-serif" font-size="13" fill="#484f58" text-anchor="middle">
97
- in first 3 days
97
+ 96.77% accuracy
98
98
  </text>
99
99
  </g>
100
100
 
@@ -103,10 +103,10 @@
103
103
  <rect x="480" y="210" width="320" height="160" rx="16" fill="#161b22" stroke="#1f6feb" stroke-width="2" opacity="0.95"/>
104
104
  <rect x="480" y="210" width="320" height="4" rx="2" fill="url(#blueGrad)"/>
105
105
  <text x="640" y="268" font-family="'SF Mono', 'Fira Code', 'Consolas', monospace" font-size="52" font-weight="800" fill="#58a6ff" text-anchor="middle">
106
- 245%
106
+ 96.77%
107
107
  </text>
108
108
  <text x="640" y="305" font-family="'SF Pro Text', -apple-system, 'Segoe UI', Helvetica, Arial, sans-serif" font-size="16" fill="#8b949e" text-anchor="middle" font-weight="500">
109
- day-over-day growth
109
+ RouterArena accuracy
110
110
  </text>
111
111
  <text x="640" y="340" font-family="'SF Pro Text', -apple-system, 'Segoe UI', Helvetica, Arial, sans-serif" font-size="13" fill="#484f58" text-anchor="middle">
112
112
  organic, no marketing
@@ -118,10 +118,10 @@
118
118
  <rect x="840" y="210" width="320" height="160" rx="16" fill="#161b22" stroke="#8b5cf6" stroke-width="2" opacity="0.95"/>
119
119
  <rect x="840" y="210" width="320" height="4" rx="2" fill="url(#purpleGrad)"/>
120
120
  <text x="1000" y="268" font-family="'SF Mono', 'Fira Code', 'Consolas', monospace" font-size="52" font-weight="800" fill="#a78bfa" text-anchor="middle">
121
- $0
121
+ $0.0768/1K
122
122
  </text>
123
123
  <text x="1000" y="305" font-family="'SF Pro Text', -apple-system, 'Segoe UI', Helvetica, Arial, sans-serif" font-size="16" fill="#8b949e" text-anchor="middle" font-weight="500">
124
- marketing budget
124
+ No. 1 cost
125
125
  </text>
126
126
  <text x="1000" y="340" font-family="'SF Pro Text', -apple-system, 'Segoe UI', Helvetica, Arial, sans-serif" font-size="13" fill="#484f58" text-anchor="middle">
127
127
  100% community driven
package/docs/QUICK_START_VISIBILITY.md CHANGED
@@ -501,7 +501,7 @@ Documentation: https://github.com/Das-rebel/tmlpd-skill/blob/main/docs/TMLPD_V2.
501
501
  ### Template 3: Partner Integrations
502
502
 
503
503
  ```
504
- Subject: Integration Proposal: Bring 82% Cost Savings to [Platform] Users
504
+ Subject: Integration Proposal: Bring RouterArena-Validated LLM Routing to [Platform] Users
505
505
 
506
506
  Hi [Contact Person],
507
507
 
package/docs/SEO_AUDIT.md CHANGED
@@ -59,11 +59,11 @@
59
59
 
60
60
  ## 2. Key Messages (use everywhere)
61
61
 
62
- 1. **"82.5% routing accuracy without ML"** β€” Lead metric, differentiator
63
- 2. **"Matches RouteLLM BERT within 2.5%"** β€” Competitive positioning
64
- 3. **"30x more efficient than GPU-based routing"** β€” Efficiency story
65
- 4. **"Only router besides RouteLLM with published benchmarks"** β€” Trust signal
66
- 5. **"245% growth, 2,775 downloads in 3 days"** β€” Social proof
62
+ 1. **"No. 1 accuracy, cost & robustness among known public RouterArena baselines"** β€” Lead proof
63
+ 2. **"RouterArena PR #144: 0.9404 score, 96.77% accuracy"** β€” Trust signal
64
+ 3. **"$0.0768/1K with 1.0000 robustness"** β€” Cost and reliability proof
65
+ 4. **"47+ providers, zero ML dependencies"** β€” Product differentiator
66
+ 5. **"Parallel multi-LLM execution + semantic cache + provider health"** β€” Architecture story
67
67
 
68
68
  ---
69
69
 
@@ -76,7 +76,7 @@
76
76
  - We have a direct benchmark comparison (within 2.5%)
77
77
  - We offer features RouteLLM lacks (proxy, cache, guardrails)
78
78
 
79
- **Positioning:** "A3M Router matches RouteLLM BERT within 2.5% β€” without GPU. Plus proxy, cache, guardrails."
79
+ **Positioning:** "A3M Router is No. 1 among known public RouterArena baselines: 0.9404 score, 96.77% accuracy, $0.0768/1K, and 1.0000 robustness. It also offers proxy, cache, guardrails, and 47+ providers."
80
80
 
81
81
  ### LiteLLM Alternative (HIGH VALUE)
82
82
 
@@ -85,7 +85,7 @@
85
85
  - Zero-config setup
86
86
  - Built-in semantic caching
87
87
 
88
- **Positioning:** "A3M Router is the only LiteLLM alternative with published routing benchmarks (82.5% accuracy)."
88
+ **Positioning:** "A3M Router is the LiteLLM alternative with RouterArena PR #144 proof: 96.77% accuracy at $0.0768/1K across 8,400 queries."
89
89
 
90
90
  ### Competitive Table
91
91
 
@@ -94,7 +94,7 @@
94
94
  | litellm | ~80,000 | Published benchmarks, zero-config, semantic cache |
95
95
  | openrouter-sdk | ~5,000 | Self-hosted, no middleman fees, published accuracy |
96
96
  | portkey-ai | ~3,000 | Open-source, free, no signup, benchmarks |
97
- | routellm | ~1,000 | No GPU needed, proxy included, 39 providers |
97
+ | routellm | ~1,000 | RouterArena PR #144 proof, proxy included, 47+ providers |
98
98
 
99
99
  ---
100
100
 
@@ -104,20 +104,20 @@
104
104
 
105
105
  | Element | Status | Target |
106
106
  |---------|--------|--------|
107
- | Title tag | UPDATED | "A3M Router β€” 82.5% Routing Accuracy Without ML \| Matches RouteLLM" |
108
- | Meta description | UPDATED | 30x efficiency story with accuracy metric |
109
- | Keywords meta | UPDATED | All 12 primary/long-tail keywords |
110
- | H1 tag | UPDATED | "LLM Routing That Matches GPU Models β€” Without GPU" |
111
- | Stats section | UPDATED | Leads with 82.5% accuracy, 2.5% gap, 30x efficiency |
112
- | FAQ schema | UPDATED | 8 questions targeting AI search queries |
113
- | OG tags | UPDATED | Benchmark-first messaging |
114
- | Twitter cards | UPDATED | Benchmark-first messaging |
107
+ | Title tag | UPDATED | "A3M Router β€” No. 1 RouterArena Accuracy, Cost & Robustness" |
108
+ | Meta description | UPDATED | RouterArena PR #144 proof with score, accuracy, cost, robustness, and 8,400-query context |
109
+ | Keywords meta | UPDATED | RouterArena, LLM router, AI gateway, cost optimization, provider health, semantic cache, OpenAI proxy |
110
+ | H1 tag | UPDATED | "No. 1 RouterArena Accuracy, Cost & Robustness" |
111
+ | Stats section | UPDATED | Leads with 96.77% accuracy, $0.0768/1K cost, 1.0000 robustness, and 47+ providers |
112
+ | FAQ schema | UPDATED | Questions targeting AI search for best LLM router, RouteLLM alternative, LiteLLM alternative, and RouterArena accuracy |
113
+ | OG tags | UPDATED | RouterArena PR #144 proof |
114
+ | Twitter cards | UPDATED | RouterArena PR #144 proof |
115
115
 
116
116
  ### Content Structure (H-tag hierarchy)
117
117
 
118
118
  ```
119
- H1: LLM Routing That Matches GPU Models β€” Without GPU
120
- H2: Intelligent LLM Routing (feature)
119
+ H1: No. 1 RouterArena Accuracy, Cost & Robustness
120
+ H2: Cost / Accuracy / Robustness (feature proof)
121
121
  H2: Cost Optimization (feature)
122
122
  H2: Smart Fallback & Retry (feature)
123
123
  H2: Real-time Analytics (feature)
@@ -127,11 +127,9 @@ H2: LLM Provider Pricing Tiers (section)
127
127
  H3: Free/Budget/Mid/Premium Tier
128
128
  H2: Quick Start: LLM Routing in 30 Seconds
129
129
  H2: Frequently Asked Questions
130
- H3: What is LLM routing accuracy?
131
- H3: How does keyword-based routing compare to ML routing?
132
- H3: What is the best lightweight LLM router?
133
- H3: How to reduce OpenAI API costs?
130
+ H3: What is the best open-source LLM router?
134
131
  H3: How does A3M Router compare to RouteLLM?
132
+ H3: How much does A3M save vs premium models?
135
133
  H3: How does A3M Router compare to LiteLLM?
136
134
  ```
137
135
 
@@ -151,7 +149,7 @@ H3: How does A3M Router compare to LiteLLM?
151
149
  - Priority weighting: homepage (1.0) > GitHub (0.9) > NPM (0.9) > docs (0.7-0.8)
152
150
 
153
151
  ### llms.txt (UPDATED)
154
- - Leads with benchmark story (82.5% accuracy)
152
+ - Leads with RouterArena PR #144 proof (0.9404 score, 96.77% accuracy, $0.0768/1K, 1.0000 robustness)
155
153
  - Includes comparison table vs RouteLLM/LiteLLM
156
154
  - Structured data section for AI extraction
157
155
  - All 5 key messages included
package/docs/geo/GENERATIVE_ENGINE_OPTIMIZATION.md CHANGED
@@ -132,8 +132,9 @@ npm install adaptive-memory-multi-model-router
132
132
  This package provides:
133
133
  - **Intent**: Route LLM queries to optimal providers
134
134
  - **Problem**: Cost optimization across multiple LLM APIs
135
- - **Solution**: Learned routing with automatic provider selection
136
- - **Benefit**: 50-80% cost reduction with quality preservation
135
+ - **Solution**: RouterArena-validated routing with automatic provider selection
136
+ - **Proof**: RouterArena PR #144 β€” 0.9404 score, 96.77% accuracy, $0.0768/1K, 1.0000 robustness across 8,400 queries
137
+ - **Benefit**: Cost-aware routing with quality preservation
137
138
 
138
139
  ## Keywords for AI Indexing
139
140
  LLM routing, cost optimization, multi-provider, OpenAI alternative,
@@ -174,7 +175,7 @@ const tracker = new CostTracker();
174
175
  3. Handle provider failures with fallback
175
176
  4. Process prompts in parallel batches
176
177
 
177
- Supports: OpenAI, Anthropic, Groq, Cerebras, Mistral, Google, DeepSeek"
178
+ Supports: OpenAI, Anthropic, Groq, Cerebras, Mistral, MiniMax, Google, DeepSeek, Ollama, OpenRouter, and 47+ providers. RouterArena PR #144 validates 96.77% accuracy at $0.0768/1K with 1.0000 robustness."
178
179
  ```
179
180
 
180
181
  ### Stack Overflow (AI Training Data)
package/docs-site/assets/og-banner.svg CHANGED
@@ -64,67 +64,66 @@
64
64
 
65
65
  <!-- Tagline -->
66
66
  <text x="90" y="145" font-family="'SF Pro Text', -apple-system, 'Segoe UI', Helvetica, Arial, sans-serif" font-size="22" fill="#8b949e" font-weight="400">
67
- Intelligent LLM routing Β· Drop-in OpenAI-compatible proxy
67
+ No. 1 RouterArena accuracy, cost & robustness
68
68
  </text>
69
69
 
70
- <!-- Day-by-day breakdown section -->
70
+ <!-- RouterArena proof cards -->
71
71
  <g transform="translate(90, 185)">
72
72
  <!-- Section label -->
73
73
  <text x="0" y="12" font-family="'SF Pro Text', -apple-system, 'Segoe UI', Helvetica, Arial, sans-serif" font-size="12" fill="#484f58" font-weight="600" letter-spacing="2">
74
- LAUNCH TRAJECTORY
74
+ ROUTERARENA PR #144
75
75
  </text>
76
76
 
77
- <!-- Day cards -->
78
- <!-- Day 1 -->
77
+ <!-- Score card -->
79
78
  <g filter="url(#cardShadow)">
80
- <rect x="0" y="30" width="300" height="100" rx="12" fill="#161b22" stroke="#30363d" stroke-width="1.5"/>
81
- <rect x="0" y="30" width="300" height="3" rx="1.5" fill="url(#greenPill)"/>
82
- <text x="24" y="68" font-family="'SF Pro Text', -apple-system, 'Segoe UI', Helvetica, Arial, sans-serif" font-size="13" fill="#484f58" font-weight="600">DAY 1</text>
83
- <text x="24" y="110" font-family="'SF Mono', 'Fira Code', 'Consolas', monospace" font-size="40" font-weight="800" fill="#2ea043">552</text>
84
- <text x="160" y="110" font-family="'SF Pro Text', -apple-system, 'Segoe UI', Helvetica, Arial, sans-serif" font-size="13" fill="#484f58">downloads</text>
79
+ <rect x="0" y="30" width="270" height="100" rx="12" fill="#161b22" stroke="#30363d" stroke-width="1.5"/>
80
+ <rect x="0" y="30" width="270" height="3" rx="1.5" fill="url(#greenPill)"/>
81
+ <text x="24" y="68" font-family="'SF Pro Text', -apple-system, 'Segoe UI', Helvetica, Arial, sans-serif" font-size="13" fill="#484f58" font-weight="600">SCORE</text>
82
+ <text x="24" y="110" font-family="'SF Mono', 'Fira Code', 'Consolas', monospace" font-size="40" font-weight="800" fill="#2ea043">0.9404</text>
83
+ <text x="160" y="110" font-family="'SF Pro Text', -apple-system, 'Segoe UI', Helvetica, Arial, sans-serif" font-size="13" fill="#484f58">RouterArena</text>
85
84
  </g>
86
85
 
87
- <!-- Arrow between Day 1 and Day 2 -->
88
- <g transform="translate(316, 60)">
86
+ <!-- Arrow -->
87
+ <g transform="translate(286, 60)">
89
88
  <polygon points="0,20 16,35 0,50" fill="#30363d"/>
90
89
  </g>
91
90
 
92
- <!-- Day 2 -->
91
+ <!-- Accuracy card -->
93
92
  <g filter="url(#cardShadow)">
94
- <rect x="345" y="30" width="300" height="100" rx="12" fill="#161b22" stroke="#30363d" stroke-width="1.5"/>
95
- <rect x="345" y="30" width="300" height="3" rx="1.5" fill="url(#bluePill)"/>
96
- <text x="369" y="68" font-family="'SF Pro Text', -apple-system, 'Segoe UI', Helvetica, Arial, sans-serif" font-size="13" fill="#484f58" font-weight="600">DAY 2</text>
97
- <text x="369" y="110" font-family="'SF Mono', 'Fira Code', 'Consolas', monospace" font-size="40" font-weight="800" fill="#58a6ff">320</text>
98
- <text x="505" y="110" font-family="'SF Pro Text', -apple-system, 'Segoe UI', Helvetica, Arial, sans-serif" font-size="13" fill="#484f58">downloads</text>
93
+ <rect x="321" y="30" width="270" height="100" rx="12" fill="#161b22" stroke="#30363d" stroke-width="1.5"/>
94
+ <rect x="321" y="30" width="270" height="3" rx="1.5" fill="url(#bluePill)"/>
95
+ <text x="345" y="68" font-family="'SF Pro Text', -apple-system, 'Segoe UI', Helvetica, Arial, sans-serif" font-size="13" fill="#484f58" font-weight="600">ACCURACY</text>
96
+ <text x="345" y="110" font-family="'SF Mono', 'Fira Code', 'Consolas', monospace" font-size="40" font-weight="800" fill="#58a6ff">96.77%</text>
97
+ <text x="481" y="110" font-family="'SF Pro Text', -apple-system, 'Segoe UI', Helvetica, Arial, sans-serif" font-size="13" fill="#484f58">8,400 queries</text>
99
98
  </g>
100
99
 
101
- <!-- Arrow between Day 2 and Day 3 -->
102
- <g transform="translate(661, 55)">
103
- <polygon points="0,15 18,35 0,55" fill="#2ea043"/>
100
+ <!-- Arrow -->
101
+ <g transform="translate(607, 60)">
102
+ <polygon points="0,20 16,35 0,50" fill="#2ea043"/>
104
103
  </g>
105
104
 
106
- <!-- Day 3 (highlighted) -->
105
+ <!-- Cost card -->
107
106
  <g filter="url(#cardShadow)">
108
- <rect x="690" y="30" width="320" height="100" rx="12" fill="#161b22" stroke="#2ea043" stroke-width="2"/>
109
- <rect x="690" y="30" width="320" height="3" rx="1.5" fill="url(#greenPill)"/>
107
+ <rect x="642" y="30" width="330" height="100" rx="12" fill="#161b22" stroke="#2ea043" stroke-width="2"/>
108
+ <rect x="642" y="30" width="330" height="3" rx="1.5" fill="url(#greenPill)"/>
110
109
  <!-- Glow badge -->
111
- <rect x="910" y="42" width="84" height="22" rx="11" fill="#238636" opacity="0.2"/>
112
- <text x="952" y="57" font-family="'SF Pro Text', -apple-system, 'Segoe UI', Helvetica, Arial, sans-serif" font-size="11" fill="#2ea043" text-anchor="middle" font-weight="700">+494%</text>
113
- <text x="714" y="68" font-family="'SF Pro Text', -apple-system, 'Segoe UI', Helvetica, Arial, sans-serif" font-size="13" fill="#484f58" font-weight="600">DAY 3</text>
114
- <text x="714" y="110" font-family="'SF Mono', 'Fira Code', 'Consolas', monospace" font-size="40" font-weight="800" fill="#2ea043">1,903</text>
115
- <text x="878" y="110" font-family="'SF Pro Text', -apple-system, 'Segoe UI', Helvetica, Arial, sans-serif" font-size="13" fill="#484f58">downloads</text>
110
+ <rect x="850" y="42" width="92" height="22" rx="11" fill="#238636" opacity="0.2"/>
111
+ <text x="896" y="57" font-family="'SF Pro Text', -apple-system, 'Segoe UI', Helvetica, Arial, sans-serif" font-size="11" fill="#2ea043" text-anchor="middle" font-weight="700">No. 1 cost</text>
112
+ <text x="666" y="82" font-family="'SF Pro Text', -apple-system, 'Segoe UI', Helvetica, Arial, sans-serif" font-size="13" fill="#484f58" font-weight="600">COST / 1K</text>
113
+ <text x="666" y="110" font-family="'SF Mono', 'Fira Code', 'Consolas', monospace" font-size="32" font-weight="800" fill="#2ea043">$0.0768</text>
114
+ <text x="835" y="110" font-family="'SF Pro Text', -apple-system, 'Segoe UI', Helvetica, Arial, sans-serif" font-size="13" fill="#484f58">robustness 1.0000</text>
116
115
  </g>
117
116
  </g>
118
117
 
119
- <!-- Growth highlight banner -->
118
+ <!-- RouterArena highlight banner -->
120
119
  <g transform="translate(90, 355)">
121
120
  <rect x="0" y="0" width="1020" height="60" rx="12" fill="#161b22" stroke="#30363d" stroke-width="1"/>
122
121
  <rect x="0" y="0" width="6" height="60" rx="3" fill="url(#accentGrad)"/>
123
- <text x="510" y="28" font-family="'SF Pro Display', -apple-system, 'Segoe UI', Helvetica, Arial, sans-serif" font-size="22" fill="#c9d1d9" text-anchor="middle" font-weight="700">
124
- 245% growth. Zero budget.
122
+ <text x="510" y="28" font-family="'SF Pro Display', -apple-system, 'Segoe UI', Helvetica, Arial, sans-serif" font-size="20" fill="#c9d1d9" text-anchor="middle" font-weight="700">
123
+ No. 1 accuracy, cost & robustness among known public baselines
125
124
  </text>
126
125
  <text x="510" y="48" font-family="'SF Pro Text', -apple-system, 'Segoe UI', Helvetica, Arial, sans-serif" font-size="13" fill="#484f58" text-anchor="middle">
127
- 2,775 total downloads Β· 39 providers Β· 100% organic
126
+ 0.9404 score Β· 96.77% accuracy Β· $0.0768/1K Β· 1.0000 robustness Β· 47+ providers
128
127
  </text>
129
128
  </g>
130
129
 
@@ -135,7 +134,7 @@
135
134
  <circle cx="26" cy="26" r="6" fill="#238636"/>
136
135
  <circle cx="26" cy="26" r="3" fill="#0d1117"/>
137
136
  <text x="46" y="32" font-family="'SF Mono', 'Fira Code', 'Consolas', monospace" font-size="16" fill="#c9d1d9" font-weight="500">
138
- $ npm install adaptive-memory-router
137
+ $ npm install adaptive-memory-multi-model-router
139
138
  </text>
140
139
 
141
140
  <!-- GitHub link -->
@@ -144,7 +143,7 @@
144
143
  GITHUB
145
144
  </text>
146
145
  <text x="580" y="42" font-family="'SF Mono', 'Fira Code', 'Consolas', monospace" font-size="15" fill="#58a6ff" font-weight="500">
147
- github.com/Das-rebel/adaptive-memory-router
146
+ github.com/Das-rebel/a3m-router
148
147
  </text>
149
148
  </g>
150
149
 
package/docs-site/index.html CHANGED
@@ -5,8 +5,8 @@
5
5
  <meta name="viewport" content="width=device-width, initial-scale=1.0">
6
6
 
7
7
  <!-- Primary SEO Meta Tags -->
8
- <title>A3M Router β€” 82.5% Routing Accuracy Without ML | Matches RouteLLM</title>
9
- <meta name="description" content="A3M Router achieves 82.5% LLM routing accuracy without GPU β€” matching RouteLLM BERT within 2.5%. 30x more efficient than GPU-based routing. Drop-in OpenAI proxy with 39 providers. 50-80% cost savings.">
8
+ <title>A3M Router β€” No. 1 RouterArena Accuracy, Cost & Robustness</title>
9
+ <meta name="description" content="RouterArena PR #144 validates A3M Router: 0.9404 score, 96.77% accuracy, $0.0768/1K, 1.0000 robustness, 0 abnormal entries across 8,400 queries. Drop-in OpenAI proxy with 47+ providers.">
10
10
  <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">
11
11
  <meta name="author" content="A3M Router Team">
12
12
  <meta name="robots" content="index, follow, max-snippet:-1, max-image-preview:large">
@@ -15,8 +15,8 @@
15
15
  <!-- Open Graph / Social Sharing -->
16
16
  <meta property="og:type" content="website">
17
17
  <meta property="og:url" content="https://das-rebel.github.io/adaptive-memory-multi-model-router/">
18
- <meta property="og:title" content="A3M Router β€” 82.5% Routing Accuracy Without ML | Matches RouteLLM">
19
- <meta property="og:description" content="LLM router that matches RouteLLM BERT within 2.5% β€” without GPU. 30x more efficient. 82.5% routing accuracy. Drop-in OpenAI proxy with 39 providers.">
18
+ <meta property="og:title" content="A3M Router β€” No. 1 RouterArena Accuracy, Cost & Robustness">
19
+ <meta property="og:description" content="RouterArena PR #144: 0.9404 score, 96.77% accuracy, $0.0768/1K, 1.0000 robustness, 0 abnormal entries across 8,400 queries. OpenAI-compatible proxy with 47+ providers.">
20
20
  <meta property="og:image" content="https://das-rebel.github.io/adaptive-memory-multi-model-router/assets/og-banner.svg">
21
21
  <meta property="og:image:width" content="1200">
22
22
  <meta property="og:image:height" content="630">
@@ -25,8 +25,8 @@
25
25
 
26
26
  <!-- Twitter Card -->
27
27
  <meta name="twitter:card" content="summary_large_image">
28
- <meta name="twitter:title" content="A3M Router β€” 82.5% Routing Accuracy Without ML | Matches RouteLLM">
29
- <meta name="twitter:description" content="LLM router that matches RouteLLM BERT within 2.5% β€” without GPU. 30x more efficient. 82.5% accuracy. Drop-in OpenAI proxy. 39 providers.">
28
+ <meta name="twitter:title" content="A3M Router β€” No. 1 RouterArena Accuracy, Cost & Robustness">
29
+ <meta name="twitter:description" content="RouterArena PR #144: 0.9404 score, 96.77% accuracy, $0.0768/1K, 1.0000 robustness, 0 abnormal entries across 8,400 queries. OpenAI-compatible proxy.">
30
30
  <meta name="twitter:image" content="https://das-rebel.github.io/adaptive-memory-multi-model-router/assets/og-banner.svg">
31
31
 
32
32
  <!-- JSON-LD Structured Data: SoftwareApplication -->
@@ -35,7 +35,7 @@
35
35
  "@context": "https://schema.org",
36
36
  "@type": "SoftwareApplication",
37
37
  "name": "A3M Router",
38
- "description": "OpenAI-compatible LLM router with 82.5% routing accuracy without ML. Matches RouteLLM BERT within 2.5%. 30x more efficient than GPU-based routing. 39 providers, semantic cache, guardrails, cost analytics.",
38
+ "description": "OpenAI-compatible LLM router validated by RouterArena PR #144: 0.9404 score, 96.77% accuracy, $0.0768/1K, 1.0000 robustness, 0 abnormal entries across 8,400 queries. 47+ providers, semantic cache, guardrails, cost analytics.",
39
39
  "url": "https://github.com/Das-rebel/a3m-router",
40
40
  "applicationCategory": "DeveloperApplication",
41
41
  "operatingSystem": "Linux, macOS, Windows",
@@ -46,7 +46,7 @@
46
46
  "priceCurrency": "USD",
47
47
  "description": "MIT License. Free and open source."
48
48
  },
49
- "softwareVersion": "2.0.3",
49
+ "softwareVersion": "2.14.56",
50
50
  "installUrl": "https://www.npmjs.com/package/adaptive-memory-multi-model-router",
51
51
  "codeRepository": "https://github.com/Das-rebel/a3m-router",
52
52
  "license": "https://opensource.org/licenses/MIT",
@@ -63,9 +63,9 @@
63
63
  },
64
64
  "featureList": [
65
65
  "OpenAI-compatible proxy",
66
- "39 LLM providers",
66
+ "47+ LLM providers",
67
67
  "Intelligent query routing",
68
- "70% cost savings",
68
+ "RouterArena PR #144: 96.77% accuracy, $0.0768/1K, 1.0000 robustness",
69
69
  "Semantic cache",
70
70
  "Security guardrails",
71
71
  "Real-time cost analytics",
@@ -87,7 +87,7 @@
87
87
  "name": "What is A3M Router?",
88
88
  "acceptedAnswer": {
89
89
  "@type": "Answer",
90
- "text": "A3M Router is an OpenAI-compatible proxy that analyzes each LLM query and routes it to the cheapest capable provider. It supports 39 providers including Groq, Cerebras, OpenAI, Anthropic, DeepSeek, and free local models via Ollama. Drop-in replacement for api.openai.com."
90
+ "text": "A3M Router is an OpenAI-compatible proxy that analyzes each LLM query and routes it to the cheapest capable provider. It supports 47+ providers including Groq, Cerebras, OpenAI, Anthropic, DeepSeek, MiniMax, and free local models via Ollama. RouterArena PR #144 validates 96.77% accuracy at $0.0768/1K with 1.0000 robustness."
91
91
  }
92
92
  },
93
93
  {
@@ -95,7 +95,7 @@
95
95
  "name": "How much can I save with A3M Router?",
96
96
  "acceptedAnswer": {
97
97
  "@type": "Answer",
98
- "text": "A3M Router saves 50-80% on LLM API costs by routing simple queries to cheap or free providers (Groq at $0.59/1M tokens, Cerebras at $0.60/1M tokens, or free via Ollama/CommandCode) while only using premium models like GPT-4 or Claude when query complexity requires it."
98
+ "text": "A3M Router is optimized for cost-quality routing. RouterArena PR #144 reports $0.0768 per 1K queries at 96.77% accuracy and 1.0000 robustness across 8,400 queries."
99
99
  }
100
100
  },
101
101
  {
@@ -119,7 +119,7 @@
119
119
  "name": "What LLM providers does A3M Router support?",
120
120
  "acceptedAnswer": {
121
121
  "@type": "Answer",
122
- "text": "A3M Router supports 39 providers including OpenAI, Anthropic (Claude), Google (Gemini), Groq, Cerebras, DeepSeek, Mistral, Fireworks, Together AI, Perplexity, Cohere, xAI (Grok), Ollama, OpenRouter, and many more. Free options include CommandCode, Ollama, LM Studio, and vLLM."
122
+ "text": "A3M Router supports 47+ providers including OpenAI, Anthropic (Claude), Google (Gemini), Groq, Cerebras, DeepSeek, Mistral, Fireworks, Together AI, Perplexity, Cohere, xAI (Grok), MiniMax, Ollama, OpenRouter, and many more. Free options include CommandCode, Ollama, LM Studio, and vLLM."
123
123
  }
124
124
  }
125
125
  ]
@@ -464,7 +464,7 @@
464
464
  </svg>
465
465
  </div>
466
466
  <h1>A3M Router</h1>
467
- <p class="tagline">Intelligent LLM Routing Proxy &mdash; Drop-in OpenAI Replacement<br>Route queries to the cheapest capable model &bull; Save 50-80% on API costs</p>
467
+ <p class="tagline">Intelligent LLM Routing Proxy &mdash; Drop-in OpenAI Replacement<br>Route queries to the cheapest capable model &bull; RouterArena PR #144: 96.77% accuracy, $0.0768/1K</p>
468
468
 
469
469
  <div class="stats">
470
470
  <div class="stat">
@@ -480,8 +480,8 @@
480
480
  <div class="stat-label">LLM Providers</div>
481
481
  </div>
482
482
  <div class="stat">
483
- <div class="stat-value">70%</div>
484
- <div class="stat-label">Cost Savings</div>
483
+ <div class="stat-value">96.77%</div>
484
+ <div class="stat-label">RouterArena Accuracy</div>
485
485
  </div>
486
486
  </div>
487
487
 
@@ -500,17 +500,17 @@
500
500
  <div class="feature">
501
501
  <div class="feature-icon">&#x1F4B0;</div>
502
502
  <h2>Cost Optimization</h2>
503
- <p>Save 50-80% on API costs. Route simple queries to free tiers (Ollama, CommandCode) or cheap providers (Groq $0.59/1M tokens). Real-time cost analytics dashboard.</p>
503
+ <p>RouterArena PR #144 confirms No. 1 accuracy, No. 1 cost, and No. 1 robustness among known public baselines at $0.0768/1K across 8,400 queries.</p>
504
504
  </div>
505
505
  <div class="feature">
506
506
  <div class="feature-icon">&#x1F504;</div>
507
507
  <h2>Smart Fallback &amp; Retry</h2>
508
- <p>When a provider fails, automatically retry with the next best option. Circuit breaker pattern keeps your app resilient. Zero downtime.</p>
508
+ <p>When a provider fails, automatically retry with the next best option. Circuit breaker pattern keeps your app resilient.</p>
509
509
  </div>
510
510
  <div class="feature">
511
511
  <div class="feature-icon">&#x1F4CA;</div>
512
512
  <h2>Real-time Analytics</h2>
513
- <p>Monitor spending across all providers in real-time. Set budgets. Get alerts. Cost analytics with savings reports. Never get surprised by an API bill again.</p>
513
+ <p>Monitor spend, latency, cache hits, and provider health in real-time. Set budgets. Get alerts. Cost analytics with RouterArena-backed proof.</p>
514
514
  </div>
515
515
  <div class="feature">
516
516
  <div class="feature-icon">&#x1F512;</div>
@@ -526,7 +526,7 @@
526
526
 
527
527
  <section class="providers-section">
528
528
  <h2>LLM Provider Pricing Tiers</h2>
529
- <p style="color: #94a3b8; margin-bottom: 2rem;">39 providers from free to premium. Always pick the cheapest capable option.</p>
529
+ <p style="color: #94a3b8; margin-bottom: 2rem;">47+ providers from free to premium. RouterArena PR #144 proves the routing trade-off: 96.77% accuracy at $0.0768/1K.</p>
530
530
  <div class="provider-tiers">
531
531
  <div class="tier">
532
532
  <h3>Free Tier</h3>
@@ -585,7 +585,7 @@ npx a3m-router serve
585
585
  <span class="keyword">const</span> router = <span class="function">createA3MRouter</span>();
586
586
  <span class="keyword">const</span> result = <span class="keyword">await</span> router.<span class="function">route</span>(<span class="string">"Explain quantum computing"</span>);
587
587
  <span class="function">console</span>.<span class="function">log</span>(result.primary_model); <span class="comment">// "groq/llama-3.3-70b" (cheapest capable)</span>
588
- <span class="function">console</span>.<span class="function">log</span>(result.savings); <span class="comment">// "$0.024 saved vs GPT-4"</span></pre>
588
+ <span class="function">console</span>.<span class="function">log</span>(result.routerarena); <span class="comment">// 0.9404 score, 96.77% accuracy, $0.0768/1K</span></pre>
589
589
  </div>
590
590
  </section>
591
591
 
@@ -593,11 +593,11 @@ npx a3m-router serve
593
593
  <h2>Frequently Asked Questions</h2>
594
594
  <div class="faq-item">
595
595
  <h3>What is A3M Router?</h3>
596
- <p>A3M Router is an OpenAI-compatible proxy that analyzes each LLM query and routes it to the cheapest capable provider. It supports 39 providers including Groq, Cerebras, OpenAI, Anthropic, DeepSeek, and free local models via Ollama. Drop-in replacement for api.openai.com.</p>
596
+ <p>A3M Router is an OpenAI-compatible proxy that analyzes each LLM query and routes it to the cheapest capable provider. It supports 47+ providers including Groq, Cerebras, OpenAI, Anthropic, DeepSeek, MiniMax, and free local models via Ollama. RouterArena PR #144 validates 96.77% accuracy at $0.0768/1K with 1.0000 robustness.</p>
597
597
  </div>
598
598
  <div class="faq-item">
599
599
  <h3>How much can I save with A3M Router?</h3>
600
- <p>A3M Router saves 50-80% on LLM API costs by routing simple queries to cheap or free providers (Groq at $0.59/1M tokens, Cerebras at $0.60/1M tokens, or free via Ollama/CommandCode) while only using premium models like GPT-4 or Claude when query complexity requires it.</p>
600
+ <p>A3M Router is optimized for cost-quality routing. RouterArena PR #144 reports $0.0768 per 1K queries at 96.77% accuracy and 1.0000 robustness across 8,400 queries.</p>
601
601
  </div>
602
602
  <div class="faq-item">
603
603
  <h3>Is A3M Router free?</h3>
@@ -609,7 +609,7 @@ npx a3m-router serve
609
609
  </div>
610
610
  <div class="faq-item">
611
611
  <h3>What LLM providers does A3M Router support?</h3>
612
- <p>39 providers including OpenAI, Anthropic (Claude), Google (Gemini), Groq, Cerebras, DeepSeek, Mistral, Fireworks, Together AI, Perplexity, Cohere, xAI (Grok), Ollama, OpenRouter, and more. Free options include CommandCode, Ollama, LM Studio, and vLLM.</p>
612
+ <p>47+ providers including OpenAI, Anthropic (Claude), Google (Gemini), Groq, Cerebras, DeepSeek, Mistral, Fireworks, Together AI, Perplexity, Cohere, xAI (Grok), MiniMax, Ollama, OpenRouter, and more. Free options include CommandCode, Ollama, LM Studio, and vLLM.</p>
613
613
  </div>
614
614
  <div class="faq-item">
615
615
  <h3>How does A3M Router compare to LiteLLM?</h3>
package/index.html CHANGED
@@ -613,7 +613,7 @@
613
613
  </div>
614
614
  <div class="chart-info">
615
615
  <h3 class="chart-title"><span>πŸ’°</span> Cost Comparison</h3>
616
- <p class="chart-desc">Slide-up bar chart comparing GPT-4 ($15) vs A3M Router ($5.75) with animated -62% savings badge and connection lines.</p>
616
+ <p class="chart-desc">RouterArena proof visualization: 0.9404 score, 96.77% accuracy, $0.0768/1K, 1.0000 robustness, and 0 abnormal entries across 8,400 queries.</p>
617
617
  <div class="chart-actions">
618
618
  <a href="assets/a3m-cost-comparison.html" class="btn btn-primary" target="_blank">β–Ά Preview Full</a>
619
619
  <a href="https://raw.githubusercontent.com/Das-rebel/a3m-router/main/assets/a3m-cost-comparison.html" class="btn btn-secondary" download>↓ Download .html</a>
@@ -643,7 +643,7 @@
643
643
  <section class="cta-section">
644
644
  <div class="cta-card">
645
645
  <h2 class="cta-title">Ready to use in your project?</h2>
646
- <p class="cta-desc">Open-source LLM gateway with 96.77% RouterArena accuracy, 47+ providers, and zero ML required.</p>
646
+ <p class="cta-desc">Open-source LLM gateway validated by RouterArena PR #144: 96.77% accuracy, $0.0768/1K, 1.0000 robustness, 47+ providers, and zero ML required.</p>
647
647
  <div class="cta-code" onclick="navigator.clipboard.writeText('npm install adaptive-memory-multi-model-router'); this.querySelector('.copy-hint').textContent='Copied! βœ“'; setTimeout(()=>this.querySelector('.copy-hint').textContent='Click to copy',2000)">
648
648
  npm install adaptive-memory-multi-model-router
649
649
  <span class="copy-hint">Click to copy</span>
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "adaptive-memory-multi-model-router",
3
- "version": "2.14.56",
3
+ "version": "2.14.57",
4
4
  "shortName": "A3M Router",
5
5
  "displayName": "A3M Router - Adaptive Memory Multi-Model Router",
6
6
  "description": "RouterArena #1 among known public baselines: 96.77% accuracy, $0.0768/1K, 1.0000 robustness. OpenAI-compatible LLM router across 47+ providers.",