wrtnlabs · samchon · Apr 8, 2026 · Apr 8, 2026 · Apr 8, 2026 · Apr 8, 2026
diff --git a/README.md b/README.md
@@ -109,10 +109,10 @@ subgraph "Functional Agents"
   coder --"Main Program" --> realize("Realize")
 end
 subgraph "Compiler Feedback"
-  database --"validates" --> prismaCompiler("<a href="https://github.com/wrtnlabs/autobe/blob/main/packages/interface/src/database/AutoBeDatabase.ts" target="_blank">Prisma Compiler</a>")
+  database --"validates" --> prismaCompiler("<a href="https://github.com/wrtnlabs/autobe/blob/main/packages/interface/src/database/AutoBeDatabase.ts" target="_blank">Database Compiler</a>")
-  database --"validates" --> prismaCompiler("<a href="https://github.com/wrtnlabs/autobe/blob/main/packages/interface/src/database/AutoBeDatabase.ts" target="_blank">Database Compiler</a>")
+  database --"validates" --> databaseCompiler("<a href="https://github.com/wrtnlabs/autobe/blob/main/packages/interface/src/database/AutoBeDatabase.ts" target="_blank">Database Compiler</a>")
-  database --"validates" --> prismaCompiler("<a href="https://github.com/wrtnlabs/autobe/blob/main/packages/interface/src/database/AutoBeDatabase.ts" target="_blank">Database Compiler</a>")
+  database --"validates" --> databaseCompiler("<a href="https://github.com/wrtnlabs/autobe/blob/main/packages/interface/src/database/AutoBeDatabase.ts" target="_blank">Database Compiler</a>")
   interface --"generates" --> openapiCompiler("<a href="https://github.com/wrtnlabs/autobe/blob/main/packages/interface/src/openapi/AutoBeOpenApi.ts" target="_blank">OpenAPI Compiler</a>")
   test --"analyzes" --> testCompiler("<a href="https://github.com/wrtnlabs/autobe/blob/main/packages/interface/src/test/AutoBeTest.ts" target="_blank">Test Compiler</a>")
-  realize --"compiles" --> realizeCompiler("TypeScript Compiler")
+  realize --"compiles" --> realizeCompiler("Hybrid Compiler")
 end
 ```
 

diff --git a/website/public/articles/autobe-qwen3.5-27b-success.html b/website/public/articles/autobe-qwen3.5-27b-success.html
@@ -0,0 +1,29 @@
+<html>
+
+<head>
+  <title>
+    [AutoBe] Qwen 3.5-27B Just Built Complete Backends from Scratch — 100% Compilation, 25x Cheaper
+  </title>
+  <meta 
+    http-equiv="refresh" 
+    content="0; url=https://dev.to/samchon/autobe-qwen-35-27b-just-built-complete-backends-from-scratch-100-compilation-25x-cheaper-lmd">
+  </meta>
-  </meta>
-  </meta>
+  <meta property="og:type" content="website">
+  <meta 
+    property="og:image"
+    content="https://autobe.dev/images/blog/qwen3.5-27b-success-og.png"
-    content="https://autobe.dev/images/blog/qwen3.5-27b-success-og.png"
+    content="https://autobe.dev/images/blog/autobe_qwen3.5-27b-success-og.png"
-    content="https://autobe.dev/images/blog/qwen3.5-27b-success-og.png"
+    content="https://autobe.dev/images/blog/autobe_qwen3.5-27b-success-og.png"
+  >
+</head>
+
+<body>
+  <p>
+    Moving to
+    <a href="https://dev.to/samchon/autobe-qwen-35-27b-just-built-complete-backends-from-scratch-100-compilation-25x-cheaper-lmd">
+      Dev.to Article
+    </a>
+    due to Reddit has shadow banned the dev.to domain links.
+  </p>
+  <p>If not redirect, click the above link manually.</p>
+</body>
+
+</html>
diff --git a/website/public/articles/autobe-vs-claude-code.html b/website/public/articles/autobe-vs-claude-code.html
@@ -11,7 +11,7 @@
   <meta property="og:type" content="website">
   <meta 
     property="og:image"
-    content="https://autobe.dev/blog/autobe-vs-claude-code-og.png"
+    content="https://autobe.dev/images/blog/autobe-vs-claude-code-og.png"
   >
 </head>
 

diff --git a/.../public/blog/autobe-vs-claude-code-og.png → .../images/blog/autobe-vs-claude-code-og.png b/.../public/blog/autobe-vs-claude-code-og.png → .../images/blog/autobe-vs-claude-code-og.png
diff --git a/website/public/images/blog/autobe_qwen3.5-27b-success-og.png b/website/public/images/blog/autobe_qwen3.5-27b-success-og.png
diff --git a/website/public/images/blog/qwen3.5-35b-a3b-legacy.png b/website/public/images/blog/qwen3.5-35b-a3b-legacy.png
diff --git a/website/public/images/demonstrate/replay-anthropic-claude-sonnet-4.6.png b/website/public/images/demonstrate/replay-anthropic-claude-sonnet-4.6.png
diff --git a/website/public/images/demonstrate/replay-minimax-minimax-m2.7.png b/website/public/images/demonstrate/replay-minimax-minimax-m2.7.png
diff --git a/website/public/images/demonstrate/replay-moonshotai-kimi-k2.5.png b/website/public/images/demonstrate/replay-moonshotai-kimi-k2.5.png
diff --git a/website/public/images/demonstrate/replay-openai-gpt-5.4-mini.png b/website/public/images/demonstrate/replay-openai-gpt-5.4-mini.png
diff --git a/website/public/images/demonstrate/replay-openai-gpt-5.4-nano.png b/website/public/images/demonstrate/replay-openai-gpt-5.4-nano.png
diff --git a/website/public/images/demonstrate/replay-openai-gpt-5.4.png b/website/public/images/demonstrate/replay-openai-gpt-5.4.png
diff --git a/website/public/images/demonstrate/replay-qwen-qwen3.5-122b-a10b.png b/website/public/images/demonstrate/replay-qwen-qwen3.5-122b-a10b.png
diff --git a/website/public/images/demonstrate/replay-qwen-qwen3.5-27b.png b/website/public/images/demonstrate/replay-qwen-qwen3.5-27b.png
diff --git a/website/public/images/demonstrate/replay-qwen-qwen3.5-35b-a3b.png b/website/public/images/demonstrate/replay-qwen-qwen3.5-35b-a3b.png
diff --git a/website/public/images/demonstrate/replay-qwen-qwen3.5-397b-a17b.png b/website/public/images/demonstrate/replay-qwen-qwen3.5-397b-a17b.png
diff --git a/website/public/images/demonstrate/replay-z-ai-glm-5.png b/website/public/images/demonstrate/replay-z-ai-glm-5.png
diff --git a/website/src/content/blog/autobe-qwen3.5-27b-success.mdx b/website/src/content/blog/autobe-qwen3.5-27b-success.mdx
@@ -0,0 +1,89 @@
+---
+title: "Qwen 3.5-27B Just Built Complete Backends from Scratch — 100% Compilation, 25x Cheaper"
+description: "Qwen 3.5-27B generated complete backends with 100% compilation at 1/25th the cost of Claude Opus 4.6 — and the output quality is nearly identical. The benchmark proves it."
+date: "2026-04-08"
+author: "Jeongho Nam"
+tags:
+  - ai
+  - typescript
+  - qwen
+  - opensource
+og: https://autobe.dev/images/blog/qwen3.5-27b-success-og.png
-og: https://autobe.dev/images/blog/qwen3.5-27b-success-og.png
+ogImage: https://autobe.dev/images/blog/autobe_qwen3.5-27b-success-og.png
-og: https://autobe.dev/images/blog/qwen3.5-27b-success-og.png
+ogImage: https://autobe.dev/images/blog/autobe_qwen3.5-27b-success-og.png
+---
+
+# Qwen 3.5-27B Just Built Complete Backends from Scratch
+
+We ran Qwen 3.5-27B on 4 backend generation tasks — from a todo app to a full ERP system. Every single project compiled. The output was nearly identical to Claude Opus 4.6, at 25x less cost.
+
+This is [AutoBe](https://github.com/wrtnlabs/autobe) — an open-source system that turns natural language into complete, compilable backend applications.
+
+![AutoBe generating a Shopping Mall backend with Qwen 3.5-27B](https://autobe.dev/images/demonstrate/replay-qwen-qwen3.5-27b.png)
+
+## 1. Generated Examples
+
+All generated by Qwen 3.5-27B. All compiled. All open source.
+
+- [Todo](https://github.com/wrtnlabs/autobe-examples/tree/main/qwen/qwen3.5-27b/todo)
+- [Reddit](https://github.com/wrtnlabs/autobe-examples/tree/main/qwen/qwen3.5-27b/reddit)
+- [Shopping](https://github.com/wrtnlabs/autobe-examples/tree/main/qwen/qwen3.5-27b/shopping)
+  - [Entity Relationship Diagram](https://github.com/wrtnlabs/autobe-examples/blob/main/qwen/qwen3.5-27b/shopping/docs/ERD.md)
+  - [API Schema](https://github.com/wrtnlabs/autobe-examples/blob/3c8bf817996a72a3bdcff791728c8dd54c3cfb4c/qwen/qwen3.5-27b/shopping/src/api/structures/IShoppingMallOrderItem.ts)
+  - [Controller](https://github.com/wrtnlabs/autobe-examples/blob/3c8bf817996a72a3bdcff791728c8dd54c3cfb4c/qwen/qwen3.5-27b/shopping/src/controllers/shoppingMall/customer/orders/ShoppingmallCustomerOrdersController.ts)
+  - [E2E Test](https://github.com/wrtnlabs/autobe-examples/blob/3c8bf817996a72a3bdcff791728c8dd54c3cfb4c/qwen/qwen3.5-27b/shopping/test/features/api/order/test_api_order_item_force_refund_single_item.ts)
+- [ERP (Enterprise Resource Planning)](https://github.com/wrtnlabs/autobe-examples/tree/main/qwen/qwen3.5-27b/erp)
+
+From a simple todo app to a full-scale ERP system. Each includes Database schema, OpenAPI spec, API implementation, E2E tests, and type-safe SDK.
+
+## 2. The Benchmark
+
+[![Benchmark: 11 AI models all scoring near-identically on backend generation](https://dev-to-uploads.s3.amazonaws.com/uploads/articles/xk2fro01vvvwm1ox4cb7.png)](https://autobe.dev/benchmark)
+
+11 models benchmarked. Scores are nearly uniform — from Qwen 3.5-27B to Claude Sonnet 4.6.
+
+A 27B model shouldn't match a frontier model. So why are the outputs identical? Because the **compiler** decides output quality — not the model.
+
+## 3. Cost
+
+| Model                     | Input / 1M tokens | Output / 1M tokens |
+|---------------------------|-------------------:|-------------------:|
+| Claude Opus 4.6           |             $5.000 |            $25.000 |
+| Qwen 3.5-27B (OpenRouter) |             $0.195 |             $1.560 |
+
+**~25x cheaper on input. ~16x on output.** Self-host Qwen and it drops to electricity.
+
+## 4. How Is This Possible?
+
+AutoBe doesn't generate text code. Instead, LLMs fill the AST structures of AutoBe's custom-built compilers through [function calling harness](https://dev.to/samchon/qwen-meetup-function-calling-harness-from-675-to-100-3830).
+
+![AutoBe's 4 compiler AST pipeline — Database, OpenAPI, Test, and Hybrid compilers validating LLM output through function calling](https://dev-to-uploads.s3.amazonaws.com/uploads/articles/dcx2zryie17ma2b2m7qx.png)
+
+Four compilers validate every output, and when something fails, the compiler's diagnoser feeds back _exactly_ what broke and why. The LLM corrects only the broken parts and resubmits — looping until every compiler passes.
+
+This harness is tight enough that model capability differences don't produce quality differences. They only affect how many retries it takes — Claude Opus gets there in 1-2 attempts, Qwen 3.5-27B in 3-4. Both converge to the same output. That's why the benchmark distribution is so uniform.
+
+> _"If you can verify, you converge."_
+
+## 5. Coming Soon: Qwen 3.5-35B-A3B
+
+![Qwen 3.5-35B-A3B benchmark showing near-complete compilation success](/images/blog/qwen3.5-35b-a3b-legacy.png)
+
+Only 3B active parameters. Not at 100% yet — but close.
+
+When it gets there: **77x cheaper**, running on a normal laptop.
+
+No cloud. No high-end GPU. Just your machine building entire backends.
+
+## 6. Try It
+
+```bash
+git clone https://github.com/wrtnlabs/autobe
+pnpm install
+pnpm playground
+```
+
+Star the repo if this is useful: **https://github.com/wrtnlabs/autobe**
+
+## 7. Deep Dives
+
+- [Function Calling Harness: From 6.75% to 100%](https://dev.to/samchon/qwen-meetup-function-calling-harness-from-675-to-100-3830)
+- [AutoBe vs. Claude Code: 3rd-Gen Coding Agent](https://dev.to/samchon/autobe-vs-claude-code-3rd-gen-coding-agent-developers-review-of-the-leaked-source-code-313b)
diff --git a/website/src/content/blog/autobe-vs-claude-code.mdx b/website/src/content/blog/autobe-vs-claude-code.mdx
@@ -2,6 +2,7 @@
 title: "AutoBE vs. Claude Code: 3rd-gen coding agent developer's review of the leaked source code"
 description: "We learned a great deal from the leaked Claude Code source code. What separates a 2nd-generation agent (Claude Code) from a 3rd-generation one (AutoBE), and where do they converge—this is a story of coexistence, not replacement."
 date: "2026-04-07"
+ogImage: "https://autobe.dev/blog/autobe-vs-claude-code-og.png"
-ogImage: "https://autobe.dev/blog/autobe-vs-claude-code-og.png"
+ogImage: "/images/blog/autobe-vs-claude-code-og.png"
-ogImage: "https://autobe.dev/blog/autobe-vs-claude-code-og.png"
+ogImage: "/images/blog/autobe-vs-claude-code-og.png"
 author: "Jeongho Nam"
 tags:
   - ai

diff --git a/website/src/content/docs/index.mdx b/website/src/content/docs/index.mdx
@@ -78,10 +78,10 @@ subgraph "Functional Agents"
   coder --"Main Program" --> realize("Realize")
 end
 subgraph "Compiler Feedback"
-  database --"validates" --> prismaCompiler("<a href="https://github.com/wrtnlabs/autobe/blob/main/packages/interface/src/database/AutoBeDatabase.ts" target="_blank">Prisma Compiler</a>")
+  database --"validates" --> prismaCompiler("<a href="https://github.com/wrtnlabs/autobe/blob/main/packages/interface/src/database/AutoBeDatabase.ts" target="_blank">Database Compiler</a>")
-  database --"validates" --> prismaCompiler("<a href="https://github.com/wrtnlabs/autobe/blob/main/packages/interface/src/database/AutoBeDatabase.ts" target="_blank">Database Compiler</a>")
+  database --"validates" --> databaseCompiler("<a href="https://github.com/wrtnlabs/autobe/blob/main/packages/interface/src/database/AutoBeDatabase.ts" target="_blank">Database Compiler</a>")
-  database --"validates" --> prismaCompiler("<a href="https://github.com/wrtnlabs/autobe/blob/main/packages/interface/src/database/AutoBeDatabase.ts" target="_blank">Database Compiler</a>")
+  database --"validates" --> databaseCompiler("<a href="https://github.com/wrtnlabs/autobe/blob/main/packages/interface/src/database/AutoBeDatabase.ts" target="_blank">Database Compiler</a>")
   interface --"generates" --> openapiCompiler("<a href="https://github.com/wrtnlabs/autobe/blob/main/packages/interface/src/openapi/AutoBeOpenApi.ts" target="_blank">OpenAPI Compiler</a>")
   test --"analyzes" --> testCompiler("<a href="https://github.com/wrtnlabs/autobe/blob/main/packages/interface/src/test/AutoBeTest.ts" target="_blank">Test Compiler</a>")
-  realize --"compiles" --> realizeCompiler("TypeScript Compiler")
+  realize --"compiles" --> realizeCompiler("Hybrid Compiler")
 end
 ```