Nexos AI Credits allow you to use large language models directly inside OpenClaw without managing individual API keys or external integrations. When OpenClaw is connected to Nexos AI, all model usage is billed through credits instead of separate provider accounts.
This makes it much easier to get started, especially if you want to experiment with different AI models or avoid setting up multiple API keys.
How credit usage is calculated
Each AI model available through Nexos AI has its own credit cost based on factors such as model size, reasoning capabilities, and response quality. More advanced models consume credits faster, while lighter models are more cost-efficient for simple tasks.
Your credit balance is shared across all OpenClaw agents connected to Nexos AI.
Supported models and pricing
The following table lists the AI models available when using Nexos AI Credits with OpenClaw, along with their credit costs.
| Model | Input cost (per 1M tokens) | Output cost (per 1M tokens) | Cache creation (per 1M tokens) | Cached input (per 1M tokens) |
|---|---|---|---|---|
| Claude Sonnet 4.5 | 3.3 credits | 16.5 credits | 4.125 credits | 0.33 credits |
| Claude Opus 4.6 | 5.5 credits | 27.5 credits | 6.75 credits | 0.55 credits |
| Claude Opus 4 | 15 credits | 75 credits | 18.75 credits | 1.5 credits |
| Gemini 3 Flash | 0.5 credits | 3 credits | Not applicable | 0.05 credits |
| GPT-4.1 | 2.2 credits | 8.8 credits | Not applicable | 0.55 credits |
| Grok 4 | 3 credits | 15 credits | Not applicable | 0.75 credits |
Choosing the right model
Each model available through Nexos AI is suited for different tasks. Here is a summary to help you choose:
Claude Sonnet 4.5 is a balanced model suited for strong reasoning, clear writing, and coding support. It works well for business tasks, structured outputs such as JSON or reports, detailed explanations, and everyday AI assistants. This is a reliable default choice for most applications that need both quality and efficiency.
Claude Opus 4.6 is a premium model built for complex analysis and advanced problem-solving. It performs well on deep research tasks, multi-step logic, long-context understanding, and high-precision outputs. Choose this model when accuracy and advanced reasoning are critical.
Claude Opus 4 is optimized for intensive reasoning and complex AI workflows. It is well-suited for enterprise-grade tasks, advanced AI agents, deep analytical work, and multi-layered problem solving. This model is ideal for demanding use cases that require maximum capability.
Gemini 3 Flash is a fast and lightweight model optimized for high-volume and quick-response tasks. It performs well for summaries, simple Q&A, content generation, and scalable chatbot applications. Choose this model when speed and responsiveness are the priority.
GPT-4.1 is a versatile model suited for structured workflows, coding, automation, and tool integration. It delivers consistent results in business environments and performs well in systems that require stable and predictable outputs. This model is a strong choice for production-ready applications.
Grok 4 is a conversational model designed for engaging interactions and idea generation. It performs well in brainstorming, social content creation, casual chat experiences, and personality-driven assistants. Choose this model for applications focused on creativity and user engagement.
What these costs mean
Input tokens are the text you send to the model, such as prompts, instructions, or conversation history.
Output tokens are the text generated by the model in response.
Cache creation applies when reusable context is stored to reduce future costs.
Cached input pricing is used when previously cached context is reused instead of being processed again.
More advanced models generally consume more credits, especially for output tokens, while lighter models like Gemini 3 Flash are more cost-efficient for simple or high-volume tasks.
Nexos AI Credits simplify the OpenClaw experience by removing manual API configuration and giving you access to multiple AI models out of the box. By adding credits during OpenClaw deployment, you can start using AI immediately and manage everything directly from the OpenClaw dashboard.