xAI Grok Models via OpenRouter

Ever Works does not include a standalone xAI plugin. Grok models from xAI are accessed through the OpenRouter plugin, which aggregates hundreds of AI models behind a single API key. This page describes how to use Grok models for work generation and AI conversations.

Related source files:

File	Purpose
packages/plugins/openrouter/src/openrouter.plugin.ts	OpenRouter AI provider plugin
apps/internal-cli/src/commands/config/ai-providers/ai-provider-registry.service.ts	Provider registry listing Grok models

Available Grok Models

The AI provider registry includes Grok models in the OpenRouter model list:

// From ai-provider-registry.service.ts
models: [
	// ...
	'x-ai/grok-3-mini-beta'
];

Grok models available through OpenRouter use the x-ai/ prefix:

Model ID	Description	Best For
x-ai/grok-3	Grok 3 full model	Complex content generation, analysis
x-ai/grok-3-mini	Grok 3 Mini	Fast, cost-effective general tasks
x-ai/grok-3-mini-beta	Grok 3 Mini Beta	Early access to latest capabilities
x-ai/grok-2	Grok 2	General purpose AI tasks

note

Model availability on OpenRouter changes over time. Check openrouter.ai/models for the current list of available Grok models and their pricing.

Configuration

Setting Up Grok Models

1. Navigate to Settings > Plugins in the Ever Works dashboard.
2. Ensure the OpenRouter plugin is enabled (default).
3. Enter your OpenRouter API key.
4. Set model fields to Grok model IDs:

Setting	Value
Default Model	x-ai/grok-3-mini
Simple Tasks Model	x-ai/grok-3-mini
Standard Tasks Model	x-ai/grok-3-mini
Complex Tasks Model	x-ai/grok-3

Environment Variables

PLUGIN_OPENROUTER_API_KEY=sk-or-...
PLUGIN_OPENROUTER_DEFAULT_MODEL=x-ai/grok-3-mini
PLUGIN_OPENROUTER_SIMPLE_MODEL=x-ai/grok-3-mini
PLUGIN_OPENROUTER_MEDIUM_MODEL=x-ai/grok-3-mini
PLUGIN_OPENROUTER_COMPLEX_MODEL=x-ai/grok-3

How It Works

Grok models are accessed through the same OpenRouter pipeline as every other provider:

OpenRouter's API is OpenAI-compatible, so the plugin uses LangChain's OpenAI provider with baseURL set to https://openrouter.ai/api/v1. The x-ai/ prefix in the model ID tells OpenRouter to route the request to xAI's infrastructure.

Tiered Model Strategy

Grok models pair well in a tiered configuration:

You can also mix Grok with other providers across tiers:

Tier	Model	Provider
Simple	openai/gpt-5-nano	OpenAI
Standard	x-ai/grok-3-mini	xAI
Complex	x-ai/grok-3	xAI

Capabilities

Capability	Grok 3	Grok 3 Mini
Structured output	Yes	Yes
Streaming	Yes	Yes
Tool calling	Yes	Yes
Vision	Model-dependent	Model-dependent
Max context length	128,000 tokens	128,000 tokens

note

Capability details depend on the specific model version available on OpenRouter. The OpenRouter plugin reports supportsVision: false at the plugin level, but individual Grok models may support vision when accessed directly.

Comparison with Other Providers

Aspect	Grok (via OpenRouter)	OpenAI (direct)	Groq
Setup	OpenRouter API key	OpenAI API key	Groq API key
Model selection	x-ai/grok-*	gpt-*, o*	Llama, Qwen
Embeddings	Depends on model	Yes	No
Speed	Standard	Standard	Ultra-fast
Cost	Varies by model	Per-token pricing	Free tier available

Reasoning Model Handling

Grok models are not currently listed in the platform's reasoning model configuration (reasoning.utils.ts). This means no automatic reasoning effort parameters are applied. If xAI introduces reasoning-specific models in the future, the platform will need a pattern entry similar to:

{
    pattern: /grok-.*-reasoning/,
    openrouter: { reasoning: { effort: 'low' } }
}

Until then, Grok models operate in standard chat completion mode.

Troubleshooting

Issue	Cause	Solution
Model not found	Incorrect model ID or model removed from OpenRouter	Verify the model ID at openrouter.ai/models
Authentication error	Invalid or expired OpenRouter API key	Regenerate your key at openrouter.ai/settings/keys
Rate limit exceeded	Too many requests to xAI via OpenRouter	Wait and retry; consider upgrading your OpenRouter plan
Embedding not supported	Grok may not offer embedding models	Use OpenAI or Google Gemini for embeddings alongside Grok
High costs	Grok full models may be expensive	Use grok-3-mini for simple and standard tiers

Cost Optimization

• Use grok-3-mini for the majority of pipeline tasks (simple and standard tiers).
• Reserve grok-3 for complex content generation only.
• Monitor usage through the OpenRouter dashboard.
• Compare per-token pricing against alternatives -- OpenRouter shows cost per model.

Further Reading

• OpenRouter Plugin -- complete OpenRouter configuration reference
• DeepSeek Models -- using DeepSeek models via OpenRouter
• AI Provider Plugins -- overview of all AI provider options
• Plugin Settings -- how plugin settings resolution works