nex-agi/Nex-N2-mini

TEXT GENERATIONConcurrency Cost:3Model Size:35.1BQuant:FP8Ctx Length:32kTool Calling:SupportedPublished:Jun 4, 2026License:apache-2.0Architecture:Transformer0.2K Open Weights Cold

The Nex-N2-mini model by nex-agi is a 35.1 billion parameter agentic language model built on the Qwen3.5-35B-A3B-Base architecture, featuring a 32768 token context length. It is specifically designed for real-world productivity scenarios, unifying reasoning, tool use, and environment execution through an "Agentic Thinking" framework. This model excels at complex, long-horizon tasks, agentic coding, deep research, and terminal execution, offering strong performance for latency and quality trade-offs.

Loading preview...

Nex-N2-mini: An Agentic Model with Agentic Thinking

Nex-N2-mini, developed by nex-agi, is a 35.1 billion parameter agentic language model built upon the Qwen3.5-35B-A3B-Base series. It is engineered for real-world productivity, integrating reasoning, tool use, and environmental execution into a unified "Agentic Thinking" framework. This framework comprises Adaptive Thinking for dynamic decision-making depth and Coherent Thinking for consistent reasoning across diverse tasks.

Key Capabilities & Features

  • Agentic Thinking Framework: Unifies requirement understanding, task planning, code implementation, environmental feedback, evaluation, debugging, and continuous iteration.
  • First-tier Performance: Achieves strong results in agentic workflows, coding tasks, and general reasoning, with substantial gains over previous generations.
  • Coding & SWE Excellence: Demonstrates robust performance on benchmarks like Terminal-Bench 2.1 (60.7) and SWE-Bench Verified (74.4).
  • Function Calling & Reasoning Parsers: Supports explicit function calling and reasoning trace parsing for enhanced control and interpretability.
  • Optimized for Productivity: Designed to handle complex, long-horizon tasks, deep research, tool calling, and terminal execution with stability and robustness.

Good For

  • Agentic Workflows: Ideal for tasks requiring autonomous execution, planning, and iteration in dynamic environments.
  • Software Engineering: Excels in code generation, debugging, and software development tasks.
  • Complex Problem Solving: Suitable for scenarios demanding thorough reasoning and multi-step task completion.
  • Real-world Productivity: A strong choice for applications needing reliable, end-to-end results in practical settings.

Finetunes
4 models