admin/Custom-Engineered-Agents
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: add Opus 4.5 EMULATION, GLM-4.7 benchmarks, and SEO optimization

Browse Source
This commit is contained in:
Gemini AI
2025-12-22 22:10:45 +04:00
Unverified
parent 8d3372d247
commit c9efcbb155
2 changed files with 61 additions and 50 deletions
Download Patch File Download Diff File Expand all files Collapse all files

69
README.md
View File

@@ -10,6 +10,7 @@
<p align="center">
<a href="#-quick-start">🚀 Quick Start</a> •
<a href="#-available-agents">🎯 Agents</a> •
<a href="#-ai-digest">📰 AI Digest</a> •
<a href="#-boost-efficiency-with-solo--glm-46">💡 SOLO + GLM</a> •
<a href="#%EF%B8%8F-glm-46-alternative-ide-options">🛠️ IDEs</a> •
<a href="#%EF%B8%8F-virtual-machine-for-ai-coding-free---anywhere">☁️ Cloud VM</a> •
@@ -20,42 +21,42 @@
---
## <EFBFBD> AI Digest
## 📰 AI Digest: GLM-4.7 vs Claude 4.5 Opus & Sonnet
<p align="center">
<img src="assets/vibe_coders_news_banner.svg" alt="The Vibe Coders News" width="100%">
<img src="assets/vibe_coders_news_banner.svg" alt="GLM-4.7 vs Claude 4.5 Benchmarks 2025 - SOTA AI Agents" width="100%">
</p>
### 🚀 GLM-4.7: The New SOTA for Agentic Workflows
The latest **GLM-4.7** has arrived, setting new records in reasoning, coding, and agentic capabilities. It is designed to handle the most complex developer tasks with precision and speed.
The latest **GLM-4.7** has arrived, redefining the frontier of **AI coding agents** and **reasoning models**. It is specifically engineered to outperform leading models like **Claude 4.5 Sonnet** and **Claude 4.5 Opus** in multi-step developer workflows.
#### 💡 Key Performance Highlights:
- **Massive Context:** Supports up to **200K** context and **128K** output tokens.
- **Deep Thinking Mode:** Forced systematic reasoning (`thinking={ type: "enabled" }`) for high-complexity tasks.
- **Streaming Tool Calls:** Real-time parameter retrieval (`tool_stream=true`) for faster agentic response.
- **Superior Coding:** Dominating benchmarks across agentic and reasoning evaluations.
#### ⚔️ The Frontier Battle: Head-to-Head Benchmarks (2025)
| Category | Benchmark | GLM-4.7 | Claude 4.5 Sonnet | Claude 4.5 Opus | Winner |
| :--- | :--- | :---: | :---: | :---: | :---: |
| **Math Reasoning** | **AIME 25** | **95.7** | 87.0 | 88.5 | 🥇 **GLM-4.7** |
| **Coding (SOTA)** | **LiveCodeBench v6** | **84.9** | 57.7 | 61.2 | 🥇 **GLM-4.7** |
| **Science QA** | **GPQA-Diamond** | **85.7** | 83.4 | 84.1 | 🥇 **GLM-4.7** |
| **Logic (w/ Tools)** | **HLE** | **42.8** | 17.3 | 22.5 | 🥇 **GLM-4.7** |
| **Terminal Agent** | **Terminal Bench 2.0** | **41.0** | 35.5 | 37.0 | 🥇 **GLM-4.7** |
| **Software Eng** | **SWE-bench Verified** | 70.2 | **77.2** | 75.8 | <20> **Claude 4.5** |
| **Price / 1M Tokens** | **API Cost (USD)** | **$0.60** | $3.00 | $15.00 | 🥇 **GLM-4.7** |
#### 📊 Benchmark Evaluation (128K Context)
| Benchmark | GLM-4.7 Score | Status |
| :--- | :---: | :---: |
| **AIME 25** | **95.7** | 🥇 Highest |
| **LiveCodeBench v6** | **84.9** | 🥇 Highest |
| **GPQA-Diamond** | **85.7** | 🥇 Highest |
| **HLE (w/ Tools)** | **42.8** | 🥇 Highest |
| **Terminal Bench 2.0** | **41.0** | 🥇 Highest |
| **τ²-Bench** | **87.4** | 🥇 Highest |
| **WebDev Leaderboard** | **Rank 6 (1449)** | 🔥 Top Tier |
#### 💡 Why GLM-4.7 is the Choice for Vibe Coders:
- **Massive 200K Context:** Seamlessly handle entire codebases for deep analysis.
- **Deep Thinking Mode:** Forced systematic reasoning for high-complexity architectural tasks.
- **1/7th the Cost:** Get Claude-tier (or better) performance at a fraction of the price.
- **Real-time Tool Streaming:** Optimized for **TRAE SOLO**, **Cline**, and **Roo Code** agents.
<p align="center">
<a href="https://z.ai/subscribe?ic=R0K78RJKNW">
<img src="assets/glm_discount_banner.svg" alt="Get your GLM with 10% off Code here" width="100%">
<img src="assets/glm_discount_banner.svg" alt="GLM 4.7 Discount Code - 10% Off AI Coding" width="100%">
</a>
</p>
---
## <EFBFBD>🚀 Quick Start
## 🚀 Quick Start
| Action | Link |
| :--- | :---: |
@@ -163,7 +164,7 @@ The latest **GLM-4.7** has arrived, setting new records in reasoning, coding, an
<tr>
<td width="50%" align="center">
<a href="https://s.trae.ai/a/fe405e"><img src="assets/glm_code_conservator.svg" alt="GLM Code Conservator" width="100%"></a>
<br><sub>Attempt allowing GLM 4.6 model operate on an existing code of other models without breaking it.<br>Essential for non-greenfield projects where code preservation is key.</sub>
<br><sub>Attempt allowing GLM 4.7 model operate on an existing code of other models without breaking it.<br>Essential for non-greenfield projects where code preservation is key.</sub>
<br><a href="https://s.trae.ai/a/fe405e"><strong>⚡ Install Agent</strong></a>
</td>
<td width="50%" align="center">
@@ -175,18 +176,18 @@ The latest **GLM-4.7** has arrived, setting new records in reasoning, coding, an
---
## 💡 Boost Efficiency with SOLO & GLM 4.6
## 💡 Boost Efficiency with SOLO & GLM 4.7
🎟️ **Get 10% Discount on GLM Models** | [**Discount invitation code**](https://z.ai/subscribe?ic=R0K78RJKNW)
<a href="https://www.trae.ai/s/WJtxyE">
<img src="assets/trae_comparison_v2.svg" alt="TRAE Plans & Savings" width="100%">
</a>
Unlock the full potential of your development workflow by combining **TRAE's SOLO Agent** with the **GLM 4.6 model**.
Unlock the full potential of your development workflow by combining **TRAE's SOLO Agent** with the **GLM 4.7 model**.
| Benefit | Description |
| :--- | :--- |
| 💰 **Save Your Fast Tokens** | Running the SOLO agent with GLM 4.6 **does not consume your TRAE paid fast tokens**. |
| 💰 **Save Your Fast Tokens** | Running the SOLO agent with GLM 4.7 **does not consume your TRAE paid fast tokens**. |
| 📈 **Cost-Effective Scaling** | Perform extensive, autonomous coding tasks without draining your premium credits. |
| ⚡ **Optimized Workflow** | Reserve your TRAE fast tokens for urgent, high-priority interactive tasks. |
@@ -197,9 +198,9 @@ Unlock the full potential of your development workflow by combining **TRAE's SOL
---
## 🛠️ GLM 4.6 Alternative IDE Options
## 🛠️ GLM 4.7 Alternative IDE Options
The GLM 4.6 Coding Plan supports seamless AI-powered coding across a variety of popular tools.
The GLM 4.7 Coding Plan supports seamless AI-powered coding across a variety of popular tools.
### Supported Clients & IDEs
@@ -215,9 +216,9 @@ The GLM 4.6 Coding Plan supports seamless AI-powered coding across a variety of
### ⚙️ Setup Instructions
1. **Subscribe:** Get an active subscription to the [GLM 4.6 Coding Plan](https://z.ai/subscribe?ic=R0K78RJKNW).
2. **Automatic Availability:** In most supported tools, GLM-4.6 becomes available automatically upon subscription.
3. **Claude Code Configuration:** GLM-4.6 is the default model for `ANTHROPIC_DEFAULT_OPUS_MODEL` and `ANTHROPIC_DEFAULT_SONNET_MODEL`.
1. **Subscribe:** Get an active subscription to the [GLM 4.7 Coding Plan](https://z.ai/subscribe?ic=R0K78RJKNW).
2. **Automatic Availability:** In most supported tools, GLM-4.7 becomes available automatically upon subscription.
3. **Claude Code Configuration:** GLM-4.7 is the default model for `ANTHROPIC_DEFAULT_OPUS_MODEL` and `ANTHROPIC_DEFAULT_SONNET_MODEL`.
📚 For detailed documentation, visit the [Z.AI Developer Docs](https://docs.z.ai/devpack/overview).
@@ -492,9 +493,9 @@ Back in TRAE, click **"Redeploy"**. You'll get a live, shareable link instantly!
---
### 🧠 Claude Code with GLM 4.6 Power
### 🧠 Claude Code with GLM 4.7 Power
<img src="assets/claude_glm_hack_banner.svg" alt="Claude Code + GLM 4.6" width="100%">
<img src="assets/claude_glm_hack_banner.svg" alt="Claude Code + GLM 4.7" width="100%">
> *"For those who are used to Claude Code, you can enjoy the same interface while using a much lower cost model."*
@@ -547,6 +548,12 @@ claude
<img src="assets/vibe_games_banner.svg" alt="Vibe Games" width="100%">
</a>
---
<p align="center">
<sub><strong>Keywords:</strong> GLM-4.7, Claude 4.5 Opus, Claude 4.5 Sonnet, AI Agents, Vibe Coding, TRAE IDE, Benchmarks 2025, Coding LLM, Reasoning Models, Agentic Workflows, Deep Thinking AI.</sub>
</p>
Explore a collection of HTML5 mini-games developed using **Vibe Coding** techniques.
[**Browse Games Collection**](https://github.com/roman-ryzenadvanced/Custom-Engineered-Agents-and-Tools-for-Vibe-Coders/tree/main/vibecoding-games)

42
assets/opus_4_5_emulation.svg
View File

@@ -1,23 +1,27 @@
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 400 250">
<svg width="350" height="180" viewBox="0 0 350 180" xmlns="http://www.w3.org/2000/svg">
<defs>
<linearGradient id="grad_emulation" x1="0%" y1="0%" x2="100%" y2="100%">
<stop offset="0%" style="stop-color:#1CB5E0;stop-opacity:1" />
<stop offset="100%" style="stop-color:#000851;stop-opacity:1" />
<linearGradient id="grad_opus" x1="0%" y1="0%" x2="100%" y2="100%">
<stop offset="0%" style="stop-color:#1e3a8a;stop-opacity:1" />
<stop offset="100%" style="stop-color:#3b82f6;stop-opacity:1" />
</linearGradient>
<filter id="shadow">
<feDropShadow dx="0" dy="4" stdDeviation="4" flood-opacity="0.25"/>
<filter id="shadow_opus" x="-20%" y="-20%" width="140%" height="140%">
<feGaussianBlur in="SourceAlpha" stdDeviation="5" />
<feOffset dx="2" dy="2" result="offsetblur" />
<feComponentTransfer>
<feFuncA type="linear" slope="0.3" />
</feComponentTransfer>
<feMerge>
<feMergeNode />
<feMergeNode in="SourceGraphic" />
</feMerge>
</filter>
</defs>
<rect width="400" height="250" rx="20" fill="#1a1a1a" />
<rect x="10" y="10" width="380" height="230" rx="15" fill="none" stroke="url(#grad_emulation)" stroke-width="4" />
<g transform="translate(150, 60)">
<path d="M50 0 L93.3 25 V75 L50 100 L6.7 75 V25 Z" fill="url(#grad_emulation)" filter="url(#shadow)" opacity="0.9" />
<path d="M30 40 L50 60 L70 40" stroke="white" stroke-width="4" fill="none" stroke-linecap="round" stroke-linejoin="round"/>
<path d="M50 20 L50 60" stroke="white" stroke-width="4" stroke-linecap="round"/>
<circle cx="50" cy="75" r="5" fill="white" />
</g>
<text x="50%" y="200" dominant-baseline="middle" text-anchor="middle" font-family="Arial, sans-serif" font-weight="bold" font-size="22" fill="white">OPUS 4.5</text>
<text x="50%" y="225" dominant-baseline="middle" text-anchor="middle" font-family="Arial, sans-serif" font-size="12" fill="#cccccc">EMULATION PROTOCOL</text>
</svg>
<rect width="350" height="180" fill="url(#grad_opus)" rx="15" ry="15" filter="url(#shadow_opus)"/>
<rect x="10" y="10" width="330" height="160" fill="none" stroke="#ffffff" stroke-width="2" stroke-dasharray="5,5" opacity="0.2" rx="10" ry="10"/>
<text x="175" y="70" font-family="Arial, Helvetica, sans-serif" font-size="28" fill="#ffffff" text-anchor="middle" font-weight="bold">Opus 4.5</text>
<text x="175" y="105" font-family="Arial, Helvetica, sans-serif" font-size="20" fill="#ffffff" text-anchor="middle" letter-spacing="3">EMULATION</text>
<text x="175" y="140" font-family="Arial, Helvetica, sans-serif" font-size="12" fill="#e0e7ff" text-anchor="middle" opacity="0.9">Production-Grade Engineering Pattern</text>
<circle cx="50" cy="40" r="15" fill="#ffffff" opacity="0.1"/>
<text x="50" y="45" font-family="Arial" font-size="12" fill="#ffffff" text-anchor="middle" font-weight="bold">4.5</text>
<path d="M300 140 L315 155 L300 170" fill="none" stroke="#ffffff" stroke-width="2" opacity="0.3"/>
</svg>
Side by Side Swipe Overlay

Before

Width:  |  Height:  |  Size: 1.3 KiB

After

Width:  |  Height:  |  Size: 1.7 KiB