adaptive-memory-multi-model-router 2.15.5 → 2.16.1

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 (206) hide show
  1. package/.github/workflows/auto-submit-sitemap.yml +41 -0
  2. package/.github/workflows/mcp-pypi-publish.yml +34 -0
  3. package/.github/workflows/pypi-publish.yml +61 -17
  4. package/.github/workflows/tmlpd-publish.yml +23 -0
  5. package/README.md +181 -207
  6. package/RELEASE_v2.16.0.md +149 -0
  7. package/TECHNICAL_README.md +253 -0
  8. package/discoverability-diagnosis.md +280 -0
  9. package/dist/analytics/costAnalytics.d.ts.map +1 -1
  10. package/dist/benchmark/reproducible.d.ts.map +1 -1
  11. package/dist/cache/semanticCache.d.ts.map +1 -1
  12. package/dist/cli/setupWizard.d.ts +257 -50
  13. package/dist/cli/setupWizard.d.ts.map +1 -1
  14. package/dist/cli/setupWizard.js +419 -109
  15. package/dist/cli/setupWizard.js.map +1 -1
  16. package/dist/cli/tui.js +96 -67
  17. package/dist/cli.js +9 -0
  18. package/dist/cost/budgetEnforcer.d.ts.map +1 -1
  19. package/dist/cost/costTracker.d.ts.map +1 -1
  20. package/dist/ensemble/multiRoundDialog.d.ts.map +1 -1
  21. package/dist/ensemble/shapleyValue.d.ts.map +1 -1
  22. package/dist/ensemble.d.ts +1 -1
  23. package/dist/ensemble.js +141 -0
  24. package/dist/integrations/langchainAdapter.d.ts.map +1 -1
  25. package/dist/integrations/langchainAdapter.js +3 -3
  26. package/dist/integrations/langchainAdapter.js.map +1 -1
  27. package/dist/integrations/oauth.d.ts.map +1 -1
  28. package/dist/integrations/scienceAdapter.d.ts.map +1 -1
  29. package/dist/memory/autoFetch.d.ts.map +1 -1
  30. package/dist/memory/hybridMemory.d.ts.map +1 -1
  31. package/dist/memory/memoryTree.d.ts.map +1 -1
  32. package/dist/memory/obsidianVault.d.ts.map +1 -1
  33. package/dist/memory/reasoningBank.d.ts.map +1 -1
  34. package/dist/observability/metrics.d.ts.map +1 -1
  35. package/dist/observability/tracer.d.ts.map +1 -1
  36. package/dist/providers/providerConfig.d.ts.map +1 -1
  37. package/dist/providers/providerConfig.js +32 -17
  38. package/dist/providers/providerConfig.js.map +1 -1
  39. package/dist/routing/advancedRouter.d.ts.map +1 -1
  40. package/dist/routing/advancedRouter.js +106 -14
  41. package/dist/routing/advancedRouter.js.map +1 -1
  42. package/dist/routing/providerHealth.d.ts.map +1 -1
  43. package/dist/routing/providerRetry.d.ts.map +1 -1
  44. package/dist/routing/shadowSampler.js.map +1 -1
  45. package/dist/security/guardrails.d.ts.map +1 -1
  46. package/dist/server/handlers/chatHandler.d.ts.map +1 -1
  47. package/dist/server/handlers/completionsHandler.d.ts.map +1 -1
  48. package/dist/server/handlers/embeddingsHandler.d.ts.map +1 -1
  49. package/dist/server/handlers/healthHandler.d.ts.map +1 -1
  50. package/dist/server/handlers/metricsHandler.d.ts.map +1 -1
  51. package/dist/server/handlers/modelsHandler.d.ts.map +1 -1
  52. package/dist/server/metrics.d.ts.map +1 -1
  53. package/dist/server/proxyServer.d.ts.map +1 -1
  54. package/dist/server/router.d.ts.map +1 -1
  55. package/dist/server/state.d.ts.map +1 -1
  56. package/dist/skills/__tests__/skill_manager.test.js +5 -265
  57. package/dist/skills/__tests__/skill_manager.test.js.map +1 -1
  58. package/dist/utils/tokenUtils.d.ts.map +1 -1
  59. package/docs/ARTICLE_Biology_Inspired_Routing.md +208 -0
  60. package/docs/ARTICLE_Master.md +78 -0
  61. package/docs/ARTICLE_Master_CN.md +78 -0
  62. package/docs/ARTICLE_OpenRouter_Stripe.md +140 -0
  63. package/docs/DEVPTO_ARTICLE.md +84 -0
  64. package/docs/HUMAN_STYLE_GUIDE.md +75 -0
  65. package/docs/IMPRINT_PLAN.md +88 -0
  66. package/docs/OPENROUTER_ALTERNATIVE.md +184 -0
  67. package/docs/SOCIAL_CAMPAIGN.md +316 -0
  68. package/docs/anthropic.html +45 -0
  69. package/docs/best-llm-routers-2025.html +157 -0
  70. package/docs/cerebras.html +43 -0
  71. package/docs/cli-cheatsheet.md +286 -212
  72. package/docs/deepseek.html +44 -0
  73. package/docs/google.html +47 -0
  74. package/docs/groq.html +44 -0
  75. package/docs/mistral.html +43 -0
  76. package/docs/ollama.html +50 -0
  77. package/docs/openai.html +57 -0
  78. package/docs/sitemap.xml +69 -57
  79. package/docs-site/blog/best-llm-routers-2025.html +157 -0
  80. package/docs-site/index.html +59 -0
  81. package/docs-site/providers/anthropic.html +45 -0
  82. package/docs-site/providers/cerebras.html +43 -0
  83. package/docs-site/providers/deepseek.html +44 -0
  84. package/docs-site/providers/google.html +47 -0
  85. package/docs-site/providers/groq.html +44 -0
  86. package/docs-site/providers/index.html +41 -0
  87. package/docs-site/providers/mistral.html +43 -0
  88. package/docs-site/providers/ollama.html +50 -0
  89. package/docs-site/providers/openai.html +57 -0
  90. package/docs-site/sitemap.xml +69 -0
  91. package/package.json +36 -124
  92. package/python/README.md +33 -106
  93. package/python/mcp-server/a3m_mcp/__init__.py +20 -10
  94. package/python/mcp-server/a3m_mcp/server.py +172 -306
  95. package/python/pyproject.toml +6 -3
  96. package/scripts/submit-sitemap.sh +52 -0
  97. package/src/__types__/registry.d.ts +14 -0
  98. package/src/cli/setupWizard.ts +443 -112
  99. package/src/cli/tui.ts +159 -0
  100. package/src/ensemble.ts +154 -1
  101. package/src/integrations/langchainAdapter.ts +2 -2
  102. package/src/providers/providerConfig.ts +32 -17
  103. package/src/providers/registry.js +27 -0
  104. package/src/routing/advancedRouter.ts +99 -14
  105. package/src/routing/shadowSampler.ts +1 -1
  106. package/test-install/package.json +12 -0
  107. package/tests/tsconfig.json +0 -1
  108. package/tmlpd-pi-extension/README.md +105 -44
  109. package/tmlpd-pi-extension/docs/demo.svg +33 -0
  110. package/tmlpd-pi-extension/package.json +35 -106
  111. package/tmlpd-pi-extension/src/tokenOptimization/contextStratifier.ts +163 -0
  112. package/tmlpd-pi-extension/src/tokenOptimization/fetchOnceLocal.ts +136 -0
  113. package/tmlpd-pi-extension/src/tokenOptimization/index.ts +197 -0
  114. package/tmlpd-pi-extension/src/tokenOptimization/interAgentCompression.ts +157 -0
  115. package/tmlpd-pi-extension/src/tokenOptimization/schemaContract.ts +101 -0
  116. package/tmlpd-pi-extension/src/tokenOptimization/semanticCache.ts +248 -0
  117. package/tmlpd-pi-extension/src/tokenOptimization/tokenAwareFallback.ts +192 -0
  118. package/tmlpd-pi-extension/test/verify.js +21 -0
  119. package/tsconfig.build.json +2 -0
  120. package/packages/a3m-vercel-ai/dist/a3m-language-model.d.ts +0 -12
  121. package/packages/a3m-vercel-ai/dist/a3m-language-model.d.ts.map +0 -1
  122. package/packages/a3m-vercel-ai/dist/a3m-language-model.js +0 -289
  123. package/packages/a3m-vercel-ai/dist/a3m-language-model.js.map +0 -1
  124. package/packages/a3m-vercel-ai/dist/index.d.ts +0 -82
  125. package/packages/a3m-vercel-ai/dist/index.d.ts.map +0 -1
  126. package/packages/a3m-vercel-ai/dist/index.js +0 -79
  127. package/packages/a3m-vercel-ai/dist/index.js.map +0 -1
  128. package/packages/a3m-vercel-ai/dist/types.d.ts +0 -97
  129. package/packages/a3m-vercel-ai/dist/types.d.ts.map +0 -1
  130. package/packages/a3m-vercel-ai/dist/types.js +0 -5
  131. package/packages/a3m-vercel-ai/dist/types.js.map +0 -1
  132. package/python/a3m_router.egg-info/PKG-INFO +0 -172
  133. package/python/a3m_router.egg-info/SOURCES.txt +0 -17
  134. package/python/a3m_router.egg-info/dependency_links.txt +0 -1
  135. package/python/a3m_router.egg-info/requires.txt +0 -24
  136. package/python/a3m_router.egg-info/top_level.txt +0 -1
  137. package/python/dist/a3m_router-2.2.1-py3-none-any.whl +0 -0
  138. package/python/dist/a3m_router-2.2.1.tar.gz +0 -0
  139. package/python/dist/a3m_router-2.2.2-py3-none-any.whl +0 -0
  140. package/python/dist/a3m_router-2.2.2.tar.gz +0 -0
  141. package/src/skills/__tests__/skill_manager.test.ts +0 -328
  142. package/tmlpd-pi-extension/dist/cache/prefixCache.d.ts +0 -114
  143. package/tmlpd-pi-extension/dist/cache/prefixCache.d.ts.map +0 -1
  144. package/tmlpd-pi-extension/dist/cache/prefixCache.js +0 -285
  145. package/tmlpd-pi-extension/dist/cache/prefixCache.js.map +0 -1
  146. package/tmlpd-pi-extension/dist/cache/responseCache.d.ts +0 -58
  147. package/tmlpd-pi-extension/dist/cache/responseCache.d.ts.map +0 -1
  148. package/tmlpd-pi-extension/dist/cache/responseCache.js +0 -153
  149. package/tmlpd-pi-extension/dist/cache/responseCache.js.map +0 -1
  150. package/tmlpd-pi-extension/dist/cli.js +0 -59
  151. package/tmlpd-pi-extension/dist/cost/costTracker.d.ts +0 -95
  152. package/tmlpd-pi-extension/dist/cost/costTracker.d.ts.map +0 -1
  153. package/tmlpd-pi-extension/dist/cost/costTracker.js +0 -240
  154. package/tmlpd-pi-extension/dist/cost/costTracker.js.map +0 -1
  155. package/tmlpd-pi-extension/dist/index.d.ts +0 -723
  156. package/tmlpd-pi-extension/dist/index.d.ts.map +0 -1
  157. package/tmlpd-pi-extension/dist/index.js +0 -239
  158. package/tmlpd-pi-extension/dist/index.js.map +0 -1
  159. package/tmlpd-pi-extension/dist/memory/episodicMemory.d.ts +0 -82
  160. package/tmlpd-pi-extension/dist/memory/episodicMemory.d.ts.map +0 -1
  161. package/tmlpd-pi-extension/dist/memory/episodicMemory.js +0 -145
  162. package/tmlpd-pi-extension/dist/memory/episodicMemory.js.map +0 -1
  163. package/tmlpd-pi-extension/dist/orchestration/haloOrchestrator.d.ts +0 -102
  164. package/tmlpd-pi-extension/dist/orchestration/haloOrchestrator.d.ts.map +0 -1
  165. package/tmlpd-pi-extension/dist/orchestration/haloOrchestrator.js +0 -207
  166. package/tmlpd-pi-extension/dist/orchestration/haloOrchestrator.js.map +0 -1
  167. package/tmlpd-pi-extension/dist/orchestration/mctsWorkflow.d.ts +0 -85
  168. package/tmlpd-pi-extension/dist/orchestration/mctsWorkflow.d.ts.map +0 -1
  169. package/tmlpd-pi-extension/dist/orchestration/mctsWorkflow.js +0 -210
  170. package/tmlpd-pi-extension/dist/orchestration/mctsWorkflow.js.map +0 -1
  171. package/tmlpd-pi-extension/dist/providers/localProvider.d.ts +0 -102
  172. package/tmlpd-pi-extension/dist/providers/localProvider.d.ts.map +0 -1
  173. package/tmlpd-pi-extension/dist/providers/localProvider.js +0 -338
  174. package/tmlpd-pi-extension/dist/providers/localProvider.js.map +0 -1
  175. package/tmlpd-pi-extension/dist/providers/registry.d.ts +0 -55
  176. package/tmlpd-pi-extension/dist/providers/registry.d.ts.map +0 -1
  177. package/tmlpd-pi-extension/dist/providers/registry.js +0 -138
  178. package/tmlpd-pi-extension/dist/providers/registry.js.map +0 -1
  179. package/tmlpd-pi-extension/dist/routing/advancedRouter.d.ts +0 -68
  180. package/tmlpd-pi-extension/dist/routing/advancedRouter.d.ts.map +0 -1
  181. package/tmlpd-pi-extension/dist/routing/advancedRouter.js +0 -332
  182. package/tmlpd-pi-extension/dist/routing/advancedRouter.js.map +0 -1
  183. package/tmlpd-pi-extension/dist/tools/tmlpdTools.d.ts +0 -101
  184. package/tmlpd-pi-extension/dist/tools/tmlpdTools.d.ts.map +0 -1
  185. package/tmlpd-pi-extension/dist/tools/tmlpdTools.js +0 -368
  186. package/tmlpd-pi-extension/dist/tools/tmlpdTools.js.map +0 -1
  187. package/tmlpd-pi-extension/dist/utils/batchProcessor.d.ts +0 -96
  188. package/tmlpd-pi-extension/dist/utils/batchProcessor.d.ts.map +0 -1
  189. package/tmlpd-pi-extension/dist/utils/batchProcessor.js +0 -170
  190. package/tmlpd-pi-extension/dist/utils/batchProcessor.js.map +0 -1
  191. package/tmlpd-pi-extension/dist/utils/compression.d.ts +0 -61
  192. package/tmlpd-pi-extension/dist/utils/compression.d.ts.map +0 -1
  193. package/tmlpd-pi-extension/dist/utils/compression.js +0 -281
  194. package/tmlpd-pi-extension/dist/utils/compression.js.map +0 -1
  195. package/tmlpd-pi-extension/dist/utils/reliability.d.ts +0 -74
  196. package/tmlpd-pi-extension/dist/utils/reliability.d.ts.map +0 -1
  197. package/tmlpd-pi-extension/dist/utils/reliability.js +0 -177
  198. package/tmlpd-pi-extension/dist/utils/reliability.js.map +0 -1
  199. package/tmlpd-pi-extension/dist/utils/speculativeDecoding.d.ts +0 -117
  200. package/tmlpd-pi-extension/dist/utils/speculativeDecoding.d.ts.map +0 -1
  201. package/tmlpd-pi-extension/dist/utils/speculativeDecoding.js +0 -246
  202. package/tmlpd-pi-extension/dist/utils/speculativeDecoding.js.map +0 -1
  203. package/tmlpd-pi-extension/dist/utils/tokenUtils.d.ts +0 -50
  204. package/tmlpd-pi-extension/dist/utils/tokenUtils.d.ts.map +0 -1
  205. package/tmlpd-pi-extension/dist/utils/tokenUtils.js +0 -124
  206. package/tmlpd-pi-extension/dist/utils/tokenUtils.js.map +0 -1
package/docs/google.html ADDED
@@ -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://das-rebel.github.io/a3m-router/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>
package/docs/groq.html ADDED
@@ -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://das-rebel.github.io/a3m-router/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>
package/docs/mistral.html ADDED
@@ -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>Mistral AI Provider — A3M Router</title>
7
+ <meta name="description" content="Configure Mistral AI (Mistral Large, Mistral 7B) with A3M Router. European AI, rate limits, and configuration.">
8
+ <meta name="robots" content="index, follow">
9
+ <link rel="canonical" href="https://das-rebel.github.io/a3m-router/providers/mistral">
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>Mistral AI Provider</h1>
16
+ <p><strong>Cost Tier:</strong> Paid · <strong>Rate Limit:</strong> Varies by tier · <strong>MCP Support:</strong> ✅ Native</p>
17
+
18
+ <h2>Supported Models</h2>
19
+ <ul>
20
+ <li><strong>Mistral Large</strong> — Flagship model for complex reasoning, coding, and analysis</li>
21
+ <li><strong>Mistral 7B</strong> — Fast, efficient open model</li>
22
+ <li><strong>Mixtral 8x22B</strong> — Mixture of experts, excellent quality</li>
23
+ <li><strong>Codestral</strong> — Specialized for code generation</li>
24
+ </ul>
25
+
26
+ <h2>How A3M Router Routes to Mistral</h2>
27
+ <p>Mistral is a strong choice for:</p>
28
+ <ul>
29
+ <li>European data residency requirements (Mistral is based in France)</li>
30
+ <li>Code generation tasks (Codestral)</li>
31
+ <li>Balanced quality/cost for general reasoning</li>
32
+ </ul>
33
+
34
+ <h2>Configuration</h2>
35
+ <pre><code>MISTRAL_API_KEY=...
36
+
37
+ # Set in environment or .env file</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>
package/docs/ollama.html ADDED
@@ -0,0 +1,50 @@
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>Ollama Provider — A3M Router</title>
7
+ <meta name="description" content="Configure Ollama (local LLM inference) with A3M Router. Zero API costs, privacy-first, and configuration guide.">
8
+ <meta name="robots" content="index, follow">
9
+ <link rel="canonical" href="https://das-rebel.github.io/a3m-router/providers/ollama">
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>Ollama Provider</h1>
16
+ <p><strong>Cost Tier:</strong> Free (hardware only) · <strong>Rate Limit:</strong> GPU-dependent · <strong>MCP Support:</strong> ✅ Native</p>
17
+
18
+ <h2>Supported Models</h2>
19
+ <ul>
20
+ <li><strong>LLaMA 3.1</strong> — Meta's open instruction-tuned model (8B, 70B)</li>
21
+ <li><strong>Mistral</strong> — Efficient open model</li>
22
+ <li><strong>Qwen 2.5</strong> — Alibaba's multilingual model</li>
23
+ <li><strong>Phi-3</strong> — Microsoft's efficient small model</li>
24
+ <li><strong>Any GGUF-format model</strong> — Custom model support</li>
25
+ </ul>
26
+
27
+ <h2>How A3M Router Routes to Ollama</h2>
28
+ <p>Ollama provides completely free, local inference. A3M Router routes to Ollama for:</p>
29
+ <ul>
30
+ <li>Maximum privacy (no data leaves your infrastructure)</li>
31
+ <li>Zero API costs for high-volume simple tasks</li>
32
+ <li>Development and testing without API key requirements</li>
33
+ </ul>
34
+
35
+ <h2>Configuration</h2>
36
+ <pre><code># Install Ollama, then pull models:
37
+ # ollama pull llama3.1
38
+ # ollama pull mistral
39
+
40
+ # A3M Router connects to local Ollama at http://localhost:11434
41
+ # No API key required — zero cost for inference</code></pre>
42
+
43
+ <h2>Requirements</h2>
44
+ <p>GPU recommended for reasonable inference speed. CPU-only works for small models (7B params) but is significantly slower.</p>
45
+
46
+ <p><a href="/providers/">← All Providers</a></p>
47
+ </main>
48
+ <footer><p>&copy; 2025 A3M Router · <a href="https://github.com/Das-rebel/a3m-router">GitHub</a></p></footer>
49
+ </body>
50
+ </html>
package/docs/openai.html ADDED
@@ -0,0 +1,57 @@
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>OpenAI Provider — A3M Router</title>
7
+ <meta name="description" content="Configure OpenAI (GPT-4o, o1, GPT-4o-mini) with A3M Router. Cost tier, rate limits, models, and configuration guide.">
8
+ <meta name="robots" content="index, follow">
9
+ <link rel="canonical" href="https://das-rebel.github.io/a3m-router/providers/openai">
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>OpenAI Provider</h1>
16
+ <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>
17
+
18
+ <h2>Supported Models</h2>
19
+ <ul>
20
+ <li><strong>GPT-4o</strong> — Latest flagship, multimodal, 128K context</li>
21
+ <li><strong>GPT-4o-mini</strong> — Cost-optimized, fast, same capabilities</li>
22
+ <li><strong>o1-preview</strong> — Reasoning model for complex tasks</li>
23
+ <li><strong>o1-mini</strong> — Reasoning model, lower cost</li>
24
+ <li><strong>GPT-4 Turbo</strong> — Previous flagship, 128K context</li>
25
+ </ul>
26
+
27
+ <h2>How A3M Router Routes to OpenAI</h2>
28
+ <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>
29
+ <ul>
30
+ <li>Complex reasoning and chain-of-thought tasks (o1 series)</li>
31
+ <li>Code generation and debugging (GPT-4o)</li>
32
+ <li>Multimodal inputs (images + text)</li>
33
+ <li>High-precision instruction following</li>
34
+ </ul>
35
+
36
+ <h2>Configuration</h2>
37
+ <pre><code># Set in your environment or .env file
38
+ OPENAI_API_KEY=sk-...
39
+
40
+ # A3M Router automatically detects and uses OpenAI when appropriate
41
+ # No additional config needed for standard use</code></pre>
42
+
43
+ <h2>Rate Limits</h2>
44
+ <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>
45
+
46
+ <h2>Cost Optimization Tips</h2>
47
+ <ul>
48
+ <li>Use <code>gpt-4o-mini</code> for simple queries — 60x cheaper than GPT-4o</li>
49
+ <li>Enable semantic caching to avoid repeat API calls</li>
50
+ <li>Route straightforward Q&amp;A to free-tier providers (Groq, Cerebras) and reserve OpenAI for complex tasks</li>
51
+ </ul>
52
+
53
+ <p><a href="/providers/">← All Providers</a></p>
54
+ </main>
55
+ <footer><p>&copy; 2025 A3M Router · <a href="https://github.com/Das-rebel/a3m-router">GitHub</a></p></footer>
56
+ </body>
57
+ </html>
package/docs/sitemap.xml CHANGED
@@ -1,57 +1,69 @@
1
- <?xml version='1.0' encoding='UTF-8'?>
2
- <ns0:urlset xmlns:ns0="http://www.sitemaps.org/schemas/sitemap/0.9">
3
- <ns0:url>
4
- <ns0:loc>https://das-rebel.github.io/a3m-router/</ns0:loc>
5
- <ns0:lastmod>2026-05-31</ns0:lastmod>
6
- <ns0:changefreq>weekly</ns0:changefreq>
7
- <ns0:priority>1.0</ns0:priority>
8
- </ns0:url>
9
- <ns0:url>
10
- <ns0:loc>https://das-rebel.github.io/a3m-router/quick-start</ns0:loc>
11
- <ns0:lastmod>2026-05-31</ns0:lastmod>
12
- <ns0:changefreq>weekly</ns0:changefreq>
13
- <ns0:priority>0.9</ns0:priority>
14
- </ns0:url>
15
- <ns0:url>
16
- <ns0:loc>https://das-rebel.github.io/a3m-router/benchmark</ns0:loc>
17
- <ns0:lastmod>2026-05-31</ns0:lastmod>
18
- <ns0:changefreq>weekly</ns0:changefreq>
19
- <ns0:priority>0.9</ns0:priority>
20
- </ns0:url>
21
- <ns0:url>
22
- <ns0:loc>https://das-rebel.github.io/a3m-router/api</ns0:loc>
23
- <ns0:lastmod>2026-05-31</ns0:lastmod>
24
- <ns0:changefreq>monthly</ns0:changefreq>
25
- <ns0:priority>0.8</ns0:priority>
26
- </ns0:url>
27
- <ns0:url>
28
- <ns0:loc>https://das-rebel.github.io/a3m-router/llms.txt</ns0:loc>
29
- <ns0:lastmod>2026-05-31</ns0:lastmod>
30
- <ns0:changefreq>weekly</ns0:changefreq>
31
- <ns0:priority>0.7</ns0:priority>
32
- </ns0:url>
33
- <ns0:url>
34
- <ns0:loc>https://das-rebel.github.io/a3m-router/llms-full.txt</ns0:loc>
35
- <ns0:lastmod>2026-05-31</ns0:lastmod>
36
- <ns0:changefreq>weekly</ns0:changefreq>
37
- <ns0:priority>0.7</ns0:priority>
38
- </ns0:url>
39
- <ns0:url>
40
- <ns0:loc>https://das-rebel.github.io/a3m-router/docs/ai-plugin.json</ns0:loc>
41
- <ns0:lastmod>2026-05-31</ns0:lastmod>
42
- <ns0:changefreq>monthly</ns0:changefreq>
43
- <ns0:priority>0.6</ns0:priority>
44
- </ns0:url>
45
- <ns0:url>
46
- <ns0:loc>https://github.com/Das-rebel/a3m-router</ns0:loc>
47
- <ns0:lastmod>2026-05-31</ns0:lastmod>
48
- <ns0:changefreq>weekly</ns0:changefreq>
49
- <ns0:priority>0.9</ns0:priority>
50
- </ns0:url>
51
- <ns0:url>
52
- <ns0:loc>https://www.npmjs.com/package/adaptive-memory-multi-model-router</ns0:loc>
53
- <ns0:lastmod>2026-05-31</ns0:lastmod>
54
- <ns0:changefreq>weekly</ns0:changefreq>
55
- <ns0:priority>0.8</ns0:priority>
56
- </ns0:url>
57
- </ns0:urlset>
1
+ <?xml version="1.0" encoding="UTF-8"?>
2
+ <urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
3
+ <url>
4
+ <loc>https://das-rebel.github.io/a3m-router/</loc>
5
+ <priority>1.0</priority>
6
+ <changefreq>weekly</changefreq>
7
+ <lastmod>2026-08-14</lastmod>
8
+ </url>
9
+ <url>
10
+ <loc>https://das-rebel.github.io/a3m-router/best-llm-routers-2025.html</loc>
11
+ <priority>0.8</priority>
12
+ <changefreq>monthly</changefreq>
13
+ <lastmod>2026-08-14</lastmod>
14
+ </url>
15
+ <url>
16
+ <loc>https://das-rebel.github.io/a3m-router/quick-start.html</loc>
17
+ <priority>0.9</priority>
18
+ <changefreq>weekly</changefreq>
19
+ <lastmod>2026-08-14</lastmod>
20
+ </url>
21
+ <url>
22
+ <loc>https://das-rebel.github.io/a3m-router/openai.html</loc>
23
+ <priority>0.6</priority>
24
+ <changefreq>monthly</changefreq>
25
+ <lastmod>2026-08-14</lastmod>
26
+ </url>
27
+ <url>
28
+ <loc>https://das-rebel.github.io/a3m-router/anthropic.html</loc>
29
+ <priority>0.6</priority>
30
+ <changefreq>monthly</changefreq>
31
+ <lastmod>2026-08-14</lastmod>
32
+ </url>
33
+ <url>
34
+ <loc>https://das-rebel.github.io/a3m-router/google.html</loc>
35
+ <priority>0.6</priority>
36
+ <changefreq>monthly</changefreq>
37
+ <lastmod>2026-08-14</lastmod>
38
+ </url>
39
+ <url>
40
+ <loc>https://das-rebel.github.io/a3m-router/groq.html</loc>
41
+ <priority>0.6</priority>
42
+ <changefreq>monthly</changefreq>
43
+ <lastmod>2026-08-14</lastmod>
44
+ </url>
45
+ <url>
46
+ <loc>https://das-rebel.github.io/a3m-router/mistral.html</loc>
47
+ <priority>0.6</priority>
48
+ <changefreq>monthly</changefreq>
49
+ <lastmod>2026-08-14</lastmod>
50
+ </url>
51
+ <url>
52
+ <loc>https://das-rebel.github.io/a3m-router/ollama.html</loc>
53
+ <priority>0.6</priority>
54
+ <changefreq>monthly</changefreq>
55
+ <lastmod>2026-08-14</lastmod>
56
+ </url>
57
+ <url>
58
+ <loc>https://das-rebel.github.io/a3m-router/deepseek.html</loc>
59
+ <priority>0.6</priority>
60
+ <changefreq>monthly</changefreq>
61
+ <lastmod>2026-08-14</lastmod>
62
+ </url>
63
+ <url>
64
+ <loc>https://das-rebel.github.io/a3m-router/cerebras.html</loc>
65
+ <priority>0.6</priority>
66
+ <changefreq>monthly</changefreq>
67
+ <lastmod>2026-08-14</lastmod>
68
+ </url>
69
+ </urlset>
package/docs-site/blog/best-llm-routers-2025.html ADDED
@@ -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>
package/docs-site/index.html CHANGED
@@ -29,6 +29,64 @@
29
29
  <meta name="twitter:description" content="Parallel multi-provider routing with confidence scoring. Semantic cache. 47+ providers. 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
+ <!-- JSON-LD Structured Data: FAQPage -->
33
+ <script type="application/ld+json">
34
+ {
35
+ "@context": "https://schema.org",
36
+ "@type": "FAQPage",
37
+ "mainEntity": [
38
+ {
39
+ "@type": "Question",
40
+ "name": "What is an LLM router?",
41
+ "acceptedAnswer": {
42
+ "@type": "Answer",
43
+ "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."
44
+ }
45
+ },
46
+ {
47
+ "@type": "Question",
48
+ "name": "How does A3M Router differ from sequential fallback?",
49
+ "acceptedAnswer": {
50
+ "@type": "Answer",
51
+ "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."
52
+ }
53
+ },
54
+ {
55
+ "@type": "Question",
56
+ "name": "Which providers does A3M Router support?",
57
+ "acceptedAnswer": {
58
+ "@type": "Answer",
59
+ "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."
60
+ }
61
+ },
62
+ {
63
+ "@type": "Question",
64
+ "name": "How fast is A3M Router?",
65
+ "acceptedAnswer": {
66
+ "@type": "Answer",
67
+ "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."
68
+ }
69
+ },
70
+ {
71
+ "@type": "Question",
72
+ "name": "Is A3M Router open source?",
73
+ "acceptedAnswer": {
74
+ "@type": "Answer",
75
+ "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."
76
+ }
77
+ },
78
+ {
79
+ "@type": "Question",
80
+ "name": "How do I get started with A3M Router?",
81
+ "acceptedAnswer": {
82
+ "@type": "Answer",
83
+ "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."
84
+ }
85
+ }
86
+ ]
87
+ }
88
+ </script>
89
+
32
90
  <!-- JSON-LD Structured Data: SoftwareApplication -->
33
91
  <script type="application/ld+json">
34
92
  {
@@ -621,6 +679,7 @@ npx a3m-router serve
621
679
  <div class="links">
622
680
  <a href="https://www.npmjs.com/package/adaptive-memory-multi-model-router">NPM</a>
623
681
  <a href="https://github.com/Das-rebel/a3m-router">GitHub</a>
682
+ <a href="/sitemap.xml">Sitemap</a>
624
683
  <a href="https://github.com/Das-rebel/a3m-router/issues">Issues</a>
625
684
  <a href="https://github.com/Das-rebel/a3m-router/discussions">Discussions</a>
626
685
  <a href="https://github.com/Das-rebel/a3m-router/blob/main/docs/API.md">API Docs</a>