Free Models

4ALL API provides a set of completely free open-source models, suitable for functionality validation, teaching demonstrations, and lightweight tasks. Free models are available under the free group; you can call them by including free in the token group.

Available Free Models in Production (Selected)

Model	Description
qwen/qwen3.6-plus:free(-preview)	Tongyi Qianwen, strong at Chinese-language tasks
z-ai/glm-4.5-air:free	Zhipu GLM, Chinese conversations
minimax/minimax-m2.5:free	MiniMax, role-based conversations
nvidia/nemotron-3-super-120b-a12b:free etc.	NVIDIA Nemotron series
google/gemma-3-27b-it:free etc.	Google Gemma open-source series
qwen/qwen3-next-80b-a3b-instruct:free	Qwen new architecture preview
stepfun/step-3.5-flash:free	StepFun fast tier
nvidia/nemotron-nano-12b-v2-vl:free	Free tier with vision capabilities

The complete list can be viewed by filtering “Free” in the Model Marketplace.

Example Call

resp = client.chat.completions.create(
    model="qwen/qwen3.6-plus:free",
    messages=[{"role": "user", "content": "Introduce yourself"}],
)

Usage Notes

• Free does not mean unlimited: concurrency and rate limits are stricter than paid tiers; see Rate Limiting and 429;
• Upstream traffic uses open-source community channels, and stability and SLA are not guaranteed. For production critical paths, please use paid models;
• Model names must include the :free suffix; models with the same name in paid tiers are different channels;
• Suitable scenarios: prototype validation, low-risk bulk coarse screening, teaching demonstrations, and baseline model comparisons.