#12020: feat: add AIsa provider, production-grade Chinese AI models
docs
commands
agents
stale
size: M
Cluster:
Model Management Enhancements
## Summary
Add **[AIsa](https://www.aisa.one/)** as a first-class model provider in OpenClaw, enabling production access to major Chinese AI models through official partnerships with leading Chinese AI platforms.
Complements the existing [Qwen Portal](https://docs.openclaw.ai/providers/qwen) (free tier, 2K req/day) by providing full model coverage, no daily caps, and up to 50% cost savings through direct partnerships.
## Official Partnerships
AIsa maintains verified partnerships with all major Chinese AI platforms:
- **Alibaba Cloud** — Qwen Key Account (full model family, 3 global regions)
- **BytePlus** — Doubao by ByteDance
- **DeepSeek** — via Alibaba Cloud integration
- **Moonshot** — Kimi integration
All partnerships can be verified upon request.
## Why AIsa?
| Feature | AIsa | Qwen Portal | OpenRouter |
| --- | --- | --- | --- |
| **Partnerships** | Official (all major Chinese AI) | OAuth free tier | Router (adds markup) |
| **Pricing** | Up to 50% off retail | Free tier | +10-25% markup |
| **Model coverage** | Qwen (all) + Doubao + DeepSeek + Kimi | 2 Qwen models | Limited Chinese models |
| **Latest models** | WAN 2.6 multimodal (20% off) | Not available | Not available |
| **Qwen regions** | CN, US (Virginia), SG via Alibaba Cloud | CN only | Global |
| **Daily limits** | None | 2,000 req/day | None |
| **Best for** | Production | Testing | Convenience |
## Pricing
AIsa provides 25-50% savings vs official Alibaba Cloud retail pricing across all Qwen tiers.
### Qwen-Flash (lightweight, high-frequency)
| Metric | AIsa | Bailian Official | OpenRouter | Savings |
| --- | --- | --- | --- | --- |
| Input/1M | **$0.05** | $0.10 | $0.11-0.13 | **50% off** |
| Output/1M | **$0.30** | $0.40 | $0.45-0.50 | **25% off** |
### Qwen-Plus (main production model)
| Metric | AIsa | Bailian Official | OpenRouter | Savings |
| --- | --- | --- | --- | --- |
| Input/1M | **$0.30** | $0.40 | $0.45-0.50 | **25% off** |
| Output/1M | **$0.90** | $1.20 | $1.35-1.50 | **25% off** |
| Thinking Output/1M | **$3.00** | $4.00 | $4.50-5.00 | **25% off** |
### Qwen-Max (flagship, complex reasoning)
| Metric | AIsa | Bailian Official | OpenRouter | Savings |
| --- | --- | --- | --- | --- |
| Input/1M | **$1.20** | $2.00 | $2.20-2.50 | **40% off** |
| Output/1M | **$4.80** | $8.00 | $9.00-10.00 | **40% off** |
### Image and Video Models (not available on OpenRouter)
| Model | AIsa | Bailian Official | OpenRouter |
| --- | --- | --- | --- |
| Qwen-VL / WAN image | **$0.02-0.03/image** | $0.04/image | N/A |
| WAN 720P video | **$0.08/sec** | $0.10/sec | N/A |
| WAN 1080P video | **$0.12/sec** | $0.15/sec | N/A |
### Cost at scale: 500M tokens/month on Qwen-Max
```
OpenRouter: $4,000-4,250/month
Bailian Official: $3,400/month
AIsa: $2,040/month (saves $16,320-26,520/year)
```
All pricing is publicly verifiable at [marketplace.aisa.one/pricing](https://marketplace.aisa.one/pricing).
## Testing
- Verified with production AIsa API key
- All 3 default models tested: qwen3-max, deepseek-v3.1, kimi-k2.5
- Model switching via `/model` command in TUI confirmed working
- Interactive onboarding flow tested
- Environment variable auth tested
- Streaming and function calling confirmed
### Screenshots
<img width="1958" alt="Screenshot from 2026-02-07 23-59-14" src="https://github.com/user-attachments/assets/16c594d1-1cb0-49c3-8ae9-23a9b7fa8930" />
<img width="5116" alt="Screenshot from 2026-02-08 00-24-41" src="https://github.com/user-attachments/assets/88ea090c-c55d-467f-96d7-cb35c8bcebc7" />
<!-- greptile_comment -->
<h2>Greptile Overview</h2>
<h3>Greptile Summary</h3>
This PR adds a new first-class model provider `aisa` (AIsa) with default models and onboarding support. Changes include:
- Docs: adds AIsa provider pages (EN + zh-CN) and links it in the providers index + docs routing (`docs/docs.json`, `docs/providers/*.md`, `docs/zh-CN/providers/*.md`).
- Runtime config: adds env var resolution for `AISA_API_KEY`, implicit provider auto-registration, and provider defaults (base URL, model list) (`src/agents/model-auth.ts`, `src/agents/models-config.providers.ts`).
- Onboarding: adds `aisa-api-key` auth choice, credential storage, config application helpers, and default model selection (`src/commands/auth-choice-options.ts`, `src/commands/auth-choice.apply.api-providers.ts`, `src/commands/onboard-auth.*`).
Integration follows existing provider patterns (env var -> auth profiles -> implicit providers; onboarding writes `auth.profiles` + `models.providers` and sets default model refs).
<h3>Confidence Score: 3/5</h3>
- This PR is mostly safe to merge, but has a couple user-facing correctness issues in onboarding/config behavior that should be fixed first.
- Core provider wiring and env/api-key resolution follow existing patterns, but the onboarding messaging is inconsistent with the provider’s actual purpose, and the model allowlisting logic is tied only to built-in defaults (can break model switching for non-default AIsa models).
- src/commands/auth-choice.apply.api-providers.ts, src/commands/onboard-auth.config-core.ts
<!-- greptile_other_comments_section -->
<sub>(3/5) Reply to the agent's comments like "Can you suggest a fix for this @greptileai?" or ask follow-up questions!</sub>
<!-- /greptile_comment -->
Most Similar PRs
#7113: feat(providers): add CommonStack provider support
by flhoildy · 2026-02-02
79.5%
#15991: feat: add Novita AI provider support with dynamic model discovery
by Alex-wuhu · 2026-02-14
77.4%
#11833: 在openclaw onboard过程中,在model/auth provider选项中增加Huawei Cloud MAAS供应商的选择
by wukunming168 · 2026-02-08
77.3%
#15327: feat:(models): discover ZAI models dynamically for GLM-5 support [A...
by vincentkoc · 2026-02-13
77.0%
#7272: Models: add SiliconFlow provider
by qychen2001 · 2026-02-02
76.8%
#2429: added cerebras as a model provider.
by kkkamur07 · 2026-01-26
76.1%
#20965: feat: Add comprehensive model configuration and discovery for various…
by rodeok · 2026-02-19
76.1%
#13079: feat: Add OpenAI-compatible API option to CLI for self-hosted models
by MikeWang0316tw · 2026-02-10
75.5%
#7295: feat(provider) Add Key-less AI Integration via Puter
by ProgrammerIn-wonderland · 2026-02-02
75.1%
#13295: feat: add Eternal AI provider integration
by peterparkernho · 2026-02-10
75.0%