DeepSeek-v3.2, Qwen, GLM, MiniMax — 8 models through one API key. No Chinese phone number needed. One free model included.
Based on moderate daily usage (~500 messages/day, mixed coding and chat tasks)
| Model | Provider | Input / 1M | Output / 1M | Context | Best For |
|---|---|---|---|---|---|
| GLM-4.6V-Flash FREE | Zhipu AI | $0 | $0 | 128K | Testing, prototyping, heartbeats |
| Qwen-Turbo | Alibaba | $0.06 | $0.20 | 128K | Classification, extraction, routing |
| DeepSeek-v3.2 POPULAR | DeepSeek | $0.20 | $0.40 | 128K | Coding, reasoning, main agent |
| Qwen-Plus | Alibaba | $0.20 | $0.60 | 128K | General purpose, multilingual |
| MiniMax-Text-01 | MiniMax | $0.20 | $1.60 | 1M | Entire repos, long documents |
| GLM-4.6V | Zhipu AI | $0.40 | $1.20 | 128K | Vision + text, multimodal |
| Qwen-Max | Alibaba | $0.40 | $1.20 | 32K | Translation, creative writing |
| Moonshot-128K | Kimi | $0.80 | $0.80 | 128K | Document analysis, summarization |
Copy this config into your ~/.openclaw/openclaw.json and you're done.
{
"models": {
"mode": "merge",
"providers": {
"novai": {
"baseUrl": "https://aiapi-pro.com/v1",
"apiKey": "YOUR_NOVAI_API_KEY",
"api": "openai-completions",
"models": [
{
"id": "deepseek-v3",
"name": "DeepSeek-v3.2",
"reasoning": true,
"input": ["text"],
"cost": { "input": 0.20, "output": 0.40 },
"contextWindow": 128000,
"maxTokens": 8192
},
{
"id": "glm-4v-flash",
"name": "GLM-4.6V-Flash (FREE)",
"reasoning": false,
"input": ["text", "image"],
"cost": { "input": 0, "output": 0 },
"contextWindow": 128000,
"maxTokens": 4096
},
{
"id": "qwen-turbo",
"name": "Qwen-Turbo",
"reasoning": false,
"input": ["text"],
"cost": { "input": 0.06, "output": 0.20 },
"contextWindow": 128000,
"maxTokens": 8192
}
]
}
}
},
// Set DeepSeek as your main model
"agents": {
"defaults": {
"model": "novai/deepseek-v3",
"models": {
"novai/deepseek-v3": { "alias": "deepseek" },
"novai/glm-4v-flash": { "alias": "glm-free" },
"novai/qwen-turbo": { "alias": "qwen" }
}
}
}
}Sign up at aiapi-pro.com with just an email. No phone verification, no credit card. Your API key appears instantly in the dashboard.
Open ~/.openclaw/openclaw.json and merge the config. Replace YOUR_NOVAI_API_KEY with your actual key. Run openclaw gateway config.apply.
Type /model novai/deepseek-v3 in your OpenClaw chat to switch. Or set it as default and forget about it. GLM-4.6V-Flash is free with no limits — great for heartbeats.
Combine these strategies to cut your bill even further.
Don't use one model for everything. Route by task complexity:
| Task | Recommended Model | Cost | Why |
|---|---|---|---|
| Heartbeats & keepalive | GLM-4.6V-Flash | $0 (FREE) | Zero cost, fast response |
| Simple lookups & classification | Qwen-Turbo | $0.06 / 1M | Cheapest paid model, very fast |
| Coding & reasoning | DeepSeek-v3.2 | $0.20 / 1M | Best cost-performance ratio |
| Full repo analysis | MiniMax-Text-01 | $0.20 / 1M | 1M context window, no chunking |
Common questions from OpenClaw users switching to NovAI.
Yes. OpenClaw supports multiple providers simultaneously. Just add NovAI as an additional provider in your config — your existing models stay untouched. Switch between them with the /model command.
On HumanEval (code benchmark), DeepSeek-v3.2 scores 90.2% vs GPT-4o's 90.2% — essentially identical. On MATH-500 it scores 90.0% vs GPT-4o's 76.6%. For most OpenClaw coding tasks, you won't notice a difference.
No. NovAI handles all upstream authentication. You sign up with email only — no phone verification, no ID check, no credit card required to start (free model available immediately).
NovAI runs from Hong Kong with direct peering to Chinese AI providers. Time to first token is under 80ms for Asia-Pacific users and around 200ms globally — comparable to or faster than US-based alternatives.
Prepaid balance, pay-as-you-go. Minimum top-up is $5 USDT. No monthly subscription, no commitments. A $5 balance typically lasts 2-4 weeks of moderate OpenClaw usage with DeepSeek.
Yes. GLM-4.6V-Flash has no usage limits and no time restriction. It's a capable 128K-context model from Zhipu AI — good for testing, heartbeats, and moderate workloads. No balance needed.
Join developers who cut their API bills by 90%. Free model included, no credit card required.
Get Your Free API Key →