Verified 2026-11-18 · rates follow the live catalog
2 models are free to call right now through the same OpenAI-compatible endpoint as every paid model. Free means free: these calls never draw down a plan allowance or a credit balance, and they work on every account, including one with no credits.
| Model id | Context | Vision | Tool calling | Last free day |
|---|---|---|---|---|
GLM-5.3-Flash-Free | 1.0M | Yes | Yes | 4 December 2026 |
DeepSeek-V4.1-Flash-1017-Free | 1.0M | Yes | Yes | 31 December 2026 |
Each free model is a limited-time offer. After its last free day the model stops being served, so do not build something permanent on one. Use it to prototype, then pick a paid model.
Create an account and an API key in the dashboard, then use the key as a bearer token. Free models answer only when streaming is on:
curl -N https://api.openference.com/v1/chat/completions \
-H "Authorization: Bearer $OPENFERENCE_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model": "GLM-5.3-Flash-Free", "stream": true,
"messages": [{"role": "user", "content": "Say hello"}]}'
Without "stream": true the request is rejected with a 400 and the code streaming_required.
Free models are governed by fair-use limits that are separate from any plan, including a cap on how many free requests can be in flight at once. A request over that cap returns a 429, and it clears once an in-progress request finishes. When the daily allowance is spent, the response is a 429 with the code daily_usage_exceeded and a resets_at timestamp saying when free access returns. Free-model limits do not apply to paid models, so a client that falls back to one keeps working.
| Model | Input / 1M | Output / 1M | One request (20,000 in, 1,000 out) |
|---|---|---|---|
| Gemma 4 26B | $0.01 | $0.03 | $0.00023 |
| GPT-OSS 20B | $0.02 | $0.03 | $0.00043 |
| Llama 4 Scout | $0.027 | $0.085 | $0.00062 |
| MiniMax M3 | $0.03 | $0.12 | $0.00072 |
| GPT-OSS 120B | $0.035 | $0.075 | $0.00077 |
The cheap AI API guide shows every billed model's cost per request, and pricing covers plans and on-demand credits.
Yes, for the models and dates listed above. They are never charged to you and never use your plan allowance or credit balance.
No. A free account with no credits can call the free models. A plan or credits are only needed for the paid models.
Any client that can point at an OpenAI-compatible base URL (https://api.openference.com/v1) can use it, as long as the client streams its responses. See the integration guides for each tool.
Model catalog · Pricing · Comparisons · Cost guides · Integration guides · FAQ
Guides: GLM coding plan alternative · Free AI API · LLM gateway · LLM proxy · Cheap AI API · OpenAI-compatible API
Model APIs: GLM API · DeepSeek API · Kimi API · Qwen API · MiniMax API