Strix now supports Vercel AI Gateway
By Steven Van ·
Strix users can set one AI Gateway key and pick from 400+ models, with usage and cost tracked in Vercel and no markup.
Strix, the open-source AI penetration testing tool, can now run on models served through Vercel AI Gateway. According to the Strix setup guide, one API key reaches models from Anthropic, OpenAI, Google, Kimi, GLM, DeepSeek and others, so separate provider sign-ups are not needed. Vercel's release note puts the Gateway catalogue at 400+ models and says there is no markup.
Setup
Create an AI Gateway API key, then set two environment variables. The vercel_ai_gateway/ prefix in STRIX_LLM selects the Gateway, and the rest of the value is the standard AI Gateway provider/model ID:
- export STRIX_LLM="vercel_ai_gateway/zai/glm-5.3"
- export LLM_API_KEY="your-ai-gateway-api-key"
Models
The Strix docs list GLM-5.3 as the recommended model. Other listed options include GPT-6.1 Sol, GPT-6 Astra, Claude Opus 5.5, Claude Sonnet 5.5, Gemini 4 Argon, Gemini 3.8 Flash, DeepSeek V4.1 Flash, DeepSeek V4 Pro, Kimi K3.1 Pro and GLM-5.3 Flash. Any model on the Gateway can be used with the vercel_ai_gateway/provider/model format. The full list is on the AI Gateway models page and at https://ai-gateway.vercel.sh/v1/models.
Safeguards and cost tracking
OpenAI, Anthropic and Gemini models keep their provider safeguards on the Gateway. Security testing with them needs Daybreak, CVP or Fairwind access. Usage and cost across all models are tracked in AI Gateway, and the Strix docs note that requests can also be routed through your own provider accounts.