Free AI Coding API: Cline, Continue, Aider and Cursor in 5 Minutes

One base URL, five free models, no credit card. Here is the exact config for each tool.

Home › Blog › Free coding API setup
Quick facts (checked 2026-10-22): 111 live models through one OpenAI-compatible API · 0% platform markup, provider list price passed through · free tier gives 10 free-model calls/day · machine-readable prices: china-ai-leaderboard.json · agent index: llms.txt

Every coding tool below speaks the OpenAI API format, so they all need the same three values:

SettingValue
Base URLhttps://aiapi-pro.com/v1
API keyyours from aiapi-pro.com/register (starts with nvai-)
Model idglm-4.7-flash (free, 202K context) or glm-4.1v-thinking-flash (free, reasoning)

Cline

  1. Open Cline settings → API Provider → OpenAI Compatible.
  2. Base URL: https://aiapi-pro.com/v1
  3. API Key: your nvai-... key
  4. Model ID: glm-4.7-flash

Cline sends tool calls and file edits; glm-4.7-flash supports function calling, so edits and diffs work normally.

Continue (VS Code / JetBrains)

Add a model entry to config.yaml:

name: My Config
version: 0.0.1
schema: v1

models:
  - name: GLM-4.7-Flash (free)
    provider: openai
    model: glm-4.7-flash
    apiKey: <YOUR_NOVAI_KEY>
    apiBase: https://aiapi-pro.com/v1/

Add a second entry with model: deepseek-v4-flash if you want a paid fallback in the same picker.

Aider

Two environment variables, or the equivalent CLI flags:

export OPENAI_API_BASE=https://aiapi-pro.com/v1
export OPENAI_API_KEY=nvai-...

aider --model openai/glm-4.7-flash

The openai/ prefix tells Aider to use the OpenAI-compatible client with your base URL rather than a built-in provider.

Cursor

  1. Settings → Models → OpenAI API Key.
  2. Paste your nvai-... key.
  3. Enable Override OpenAI Base URL and set https://aiapi-pro.com/v1.
  4. Add a custom model name: glm-4.7-flash.

Which free model for which job

ModelContextUse it for
glm-4.7-flash202KEveryday edits, refactors, multi-file changes
glm-4.1v-thinking-flash128KHarder bugs where you want visible reasoning

Both are free to call: 10 free-model calls per day and 5 requests per minute on the free tier, no card required. That is enough to evaluate a tool end-to-end, not enough for a full day of agent work.

When the free cap is not enough: requests return 429 until 00:00 UTC+8. A one-time top-up removes the cap and raises the rate limit to 60 req/min; the first top-up carries a 25% bonus. Paid coding models start at $0.08 per 1M input tokens — see cheapest model per task.

FAQ

Which free model should I use for coding?

GLM-4.7-Flash (202K context) for general coding and edits; GLM-4.1V-Thinking-Flash (128K) when you want visible reasoning steps for harder problems. Both are free to call, capped at 10 free-model calls per day on the free tier.

Does the free tier work with Cline, Continue, Aider and Cursor?

Yes. All four support an OpenAI-compatible endpoint. Point the base URL at https://aiapi-pro.com/v1, use an nvai- key, and pick one of the five free model ids.

What happens when the daily free cap is reached?

Requests to free models return HTTP 429 until 00:00 UTC+8. A one-time top-up removes the free-tier cap and raises the rate limit to 60 requests per minute; the first top-up carries a 25% bonus.

Can I use paid models in the same tool config?

Yes. Add more entries with paid model ids (for example deepseek-v4-flash or kimi-k2.7-code) using the same base URL and key. Paid models are billed per token at provider list price with no platform markup.

Five free models, one key

$2 credit on signup, no card required. 111 models behind the same endpoint when you outgrow the free tier.

Get a free API key

Setup steps follow each tool's current OpenAI-compatible configuration. Limits and prices reflect the live catalog on 2026-10-22; verify at /transparency.html and /china-ai-leaderboard.json.