How do I use Aider with an OpenAI-compatible gateway?

Use the openai/ model prefix with --openai-api-base https://aiapi-pro.com/v1 and --openai-api-key. Example: aider --model openai/glm-4.7-flash --openai-api-base https://aiapi-pro.com/v1 --openai-api-key $NOVAI_API_KEY

Which Chinese model is best for Aider coding tasks?

For hard refactors use deepseek-v4-pro or kimi-k2.7-code; for fast everyday edits qwen3-coder-flash or glm-5.3-flash; for zero cost use the free unlimited glm-4.7-flash.

Is there a platform markup on NovAI prices?

No - 0% platform fee, raw provider list price (vs OpenRouter ~5.5%). Live prices at /china-ai-leaderboard.json.

Aider + Chinese AI Models (2026)

Updated 2026-10-10

Aider is the terminal pair programmer that edits your repo directly. Because it accepts any OpenAI-compatible endpoint, you can drive it with Chinese frontier models — including ones that cost nothing — through NovAI.

One-line setup

pip install aider-install && aider-install

export NOVAI_API_KEY=sk-...   # https://aiapi-pro.com/register ($2 free)
aider --model openai/glm-4.7-flash \
      --openai-api-base https://aiapi-pro.com/v1 \
      --openai-api-key $NOVAI_API_KEY

The openai/ prefix tells Aider to use the OpenAI chat protocol against your base URL; the model list (121 models) is whatever id you pass.

Persistent config (.aider.conf.yml)

model: openai/kimi-k2.7-code
openai-api-base: https://aiapi-pro.com/v1
openai-api-key: $NOVAI_API_KEY

Model picks by task

Task	Model id	Cost
Zero-cost everyday edits	glm-4.7-flash	Free, unlimited
Fast code completion	qwen3-coder-flash	see leaderboard
Balanced refactors	glm-5.3-flash	$0.08 / $0.28 per 1M
Hard architecture work	deepseek-v4-pro / kimi-k2.7-code	see leaderboard

Free unlimited models right now: GLM-4.7-Flash, GLM-4.6V-Flash, GLM-4.1V-Thinking-Flash, CogView3-Flash, CogVideoX-Flash. Live per-model pricing: leaderboard / JSON. NovAI adds 0% platform fee.

Tips

Start sessions on a free model, then /model openai/deepseek-v4-pro to escalate only for hard steps — cuts cost dramatically.
Use --map-tokens defaults; repo-map calls are cheap on flash models.
If Aider reports a model not found, check the exact id at the catalog or curl https://aiapi-pro.com/v1/models.

Get $2 free credit · Live price leaderboard · API docs

Related: Claude Code · Cline & Roo Code · Cursor · Open WebUI