1. Highest reasoning quality in the GPT-5 family

• Uses significantly more compute to "think harder" before responding.
• Designed for the toughest reasoning tasks where answer quality matters more than speed.
• Produces more precise, reliable, and detailed outputs than standard GPT-5.

2. Advanced multi-turn reasoning via Responses API

• Available only in the Responses API to support:
  • Multi-turn internal model interactions before returning a reply.
  • Advanced control patterns (e.g., background mode for long-running jobs).
• Ideal for complex workflows, deep planning, and multi-step analysis.

3. Configured for maximum effort by default

• Always runs with reasoning.effort: 'high' (no lower-effort mode).
• Prioritizes depth and correctness over latency and cost.

4. Multimodal input

• Accepts text + image as input.
• Outputs text, with strong instruction-following and analysis capabilities.

5. Tooling and ecosystem integration

• Supports Web Search, File Search, and Image Generation (as tools).
• Supports MCP and other Responses API tooling patterns.
• Does not support Code Interpreter and does not support Computer Use, keeping focus on pure reasoning + tools.

• Open-weight / Apache 2.0 licensed: you can use, modify, and deploy freely (commercially & academically) under permissive terms.
• Large model size (≈ 21B parameters) with Mixture-of-Experts (MoE) architecture: only ~3.6B parameters active per token, yielding efficient inference.
• Very long context window support: up to ~128 K tokens (or ~131 K tokens per some sources) enabling in-depth reasoning, long documents, or multi-turn context.
• Adjustable reasoning effort: you can trade latency vs quality by tuning “reasoning effort” levels.
• Efficient hardware requirements (for its class): designed to run on a single 16 GB-class GPU or optimized local deployments for lower latency applications.
• Strong for tasks such as reasoning, tool-use, structured output, chain-of-thought debugging: because the model is open and you can inspect its chain of thought.
• Flexibility: since weights are available, you can self-host, fine-tune, or deploy offline, giving more control than closed API models.