GPT-6 Sol is now the recommended model for prompt-based Conversation AI agents. We optimized Conversation AI to get the best out of GPT-6 Sol, so your agents give more accurate answers than on GPT-4.1, at about the same cost per reply.
Key Capabilities
- Recommended in the Model menu: GPT-6 Sol is listed first in the Model menu and tagged Recommended.
- New agents start on GPT-6 Sol: Prompt-based agents you create now use GPT-6 Sol by default.
- Your choice stays: Existing agents keep the model you selected.
- More accurate at the same cost: GPT-6 Sol reaches a 98.4% pass rate on our Conversation AI evaluation, up from 96.3% on GPT-4.1, at about the same cost per reply.
- Fast replies: 1.68s typical generation time, and 2.19s for the slowest 5% of replies.
Changes to Older Models
- GPT-5 and GPT-5 Mini are deprecated: Agents that used them already run on GPT-6 Sol, with reply generation up to 4.5× faster. No action needed.
- GPT-4.1 is being retired: OpenAI is deprecating GPT-4.1, and we will retire it in Conversation AI. Switch your GPT-4.1 agents to GPT-6 Sol.
How to Get Started
- Go to AI Agents → Conversation AI and open your agent.
- Open the Model menu on the Build tab.
- Select GPT 6 Sol (marked Recommended).
- Click Save, then try it in Test your agent.
Why It Matters
- Better answers out of the box: Every new agent starts on the most accurate model we tested.
- No extra cost: Higher accuracy than GPT-4.1 at about the same cost per reply.
- Faster replies: Agents that were on GPT-5 and GPT-5 Mini now generate replies up to 4.5× faster.
- An easy move off GPT-4.1: Switch before GPT-4.1 is retired.
Important Notes
- Test after you switch: Each model writes in its own way, so test your agent in Test your agent before you save.
- Prompt-based agents: The recommendation applies to prompt-based Conversation AI agents.
