AarthShah/Nex-N2-mini

TEXT GENERATIONPricing:Input $0.4 / Cached $0.07 / Output $4Concurrent Unit Cost:2Model Size:35.1BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Sep 28, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

Nex-N2-mini by Nex-AGI is a 35.1 billion parameter agentic language model, built on the Qwen3.5-35B-A3B-Base series, designed for real-world productivity scenarios. It unifies reasoning, tool use, and environmental execution through an "Agentic Thinking" framework, excelling in complex, long-horizon tasks. This model demonstrates strong performance in agentic coding, deep research, and terminal execution, offering a balance between latency and quality.

Loading preview...

Nex-N2-mini: An Agentic Model for Real-World Productivity

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 specifically engineered for real-world productivity, focusing on driving complex, long-horizon tasks with stable, end-to-end results. The model integrates reasoning, tool use, and environmental execution through an innovative Agentic Thinking framework.

Key Capabilities

  • Agentic Thinking Framework: Unifies requirement understanding, task planning, code implementation, environmental feedback, evaluation, debugging, and continuous iteration into a single closed loop.
  • Adaptive Thinking: Allows the model to dynamically adjust its reasoning depth, executing simple actions quickly while thoroughly reasoning on critical decisions.
  • Coherent Thinking: Maintains a consistent reasoning paradigm across general and diverse agentic tasks, ensuring stable capability transfer.
  • Strong Performance: Achieves competitive scores in agentic workflows, coding tasks, and general reasoning, including 60.7 on Terminal-Bench 2.1 and 1402 on GDPval.
  • Function Calling & Reasoning Traces: Supports robust function-calling and emits explicit reasoning traces for enhanced transparency and control.

Good For

  • Agentic Coding: Excels in software engineering tasks and terminal execution.
  • Deep Research: Capable of handling complex information retrieval and synthesis.
  • Tool Calling: Designed for effective interaction with external tools and environments.
  • Long-Horizon Tasks: Ideal for scenarios requiring sustained, multi-step problem-solving and iteration.

Finetunes
1 models