EdmondMillion/affine-7-5EXDeevNLXBeWscrMYoCs9eNmfxiEd5tzSeR3DxkoDsZkiy7
TEXT GENERATIONConcurrency Cost:1Model Size:4BQuant:BF16Ctx Length:32kPublished:Feb 2, 2026Architecture:Transformer Cold

The EdmondMillion/affine-7-5EXDeevNLXBeWscrMYoCs9eNmfxiEd5tzSeR3DxkoDsZkiy7 model is part of the GLM-4-32B-0414 series, a 32 billion parameter model family developed by GLM. This series, including GLM-Z1-32B-0414 for deep reasoning and GLM-Z1-Rumination-32B-0414 for complex, open-ended problem-solving with search tools, is pre-trained on 15T high-quality data. It excels in engineering code, artifact generation, function calling, search-based Q&A, and report generation, with performance comparable to larger models like GPT-4o and DeepSeek-V3-0324 on specific benchmarks.

Loading preview...

Overview

This model, part of the GLM-4-32B-0414 series, is a 32 billion parameter language model developed by GLM. It was pre-trained on 15 trillion tokens of high-quality data, including significant reasoning-type synthetic data, and further enhanced with human preference alignment and reinforcement learning. The series includes specialized variants like GLM-Z1-32B-0414 for deep reasoning, with improved mathematical and complex task-solving abilities, and GLM-Z1-Rumination-32B-0414, designed for deeper, longer thinking on open-ended problems, capable of using search tools during its process. A smaller 9B parameter model, GLM-Z1-9B-0414, also demonstrates strong mathematical reasoning and general task performance, making it suitable for resource-constrained environments.

Key Capabilities

  • Engineering Code & Artifact Generation: Achieves strong results in generating code and various artifacts.
  • Function Calling: Supports external tool calls using a JSON message format, demonstrated with Python examples.
  • Search-Based Q&A and Report Generation: Excels at generating detailed analytical reports based on provided search results.
  • Deep Reasoning: GLM-Z1-32B-0414 significantly improves mathematical abilities and complex task-solving.
  • Rumination: GLM-Z1-Rumination-32B-0414 offers deeper, longer thinking for open-ended problems, leveraging search tools.

Performance Highlights

  • On benchmarks like IFEval, BFCL-v3, TAU-Bench, SimpleQA, and HotpotQA, GLM-4-32B-0414 shows competitive performance, often surpassing or matching larger models like GPT-4o and DeepSeek-V3-0324 in specific areas.
  • Demonstrates strong performance on SWE-bench Verified and SWE-bench Verified mini using various agent frameworks.

Good For

  • Developers requiring robust code generation and function calling capabilities.
  • Applications needing advanced reasoning, especially in mathematics and complex problem-solving.
  • Tasks involving search-augmented content generation, such as detailed Q&A and report writing.
  • Scenarios where a powerful yet efficient model is needed, particularly the 9B variant for lightweight deployment.