How do I switch between models on the gateway?

Models are addressed as vendor/model, such as deepseek/deepseek-v4-flash or anthropic/claude-opus-4-8, so switching is a string change rather than a migration.

Do I need a Chinese phone number or payment method to reach Chinese models?

No, according to AICraft. You sign up once, in English, and the whole catalogue, including Chinese and Western models, becomes reachable through the same endpoint.

When is a direct provider call simpler than using a gateway?

If you only ever use a single Western model and deploy in one region, calling that provider directly is the simpler answer, and a gateway adds a hop for no benefit.

Large Language ModelsAI Coding AssistantsZenn AI/ML1d agoOct 14, 2026, 22:00 JST

AICraft opens one OpenAI-compatible key to 100+ models

Send on LINE

3 Key Points

What happened
AICraft launched an OpenAI-compatible API gateway at https://aicraftapi.net/v1 that reaches 100+ models under one key, base URL, and invoice, replacing per-provider accounts and billing dashboards.
Why it matters
The gateway removes the multiple sign-ups, separate billing dashboards, and integration rewrites that teams face when they build with language models across several providers, according to AICraft.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Get news like this every morning, free

Context & Analysis

Teams that build with language models often end up maintaining a separate account, API key, billing dashboard, and integration code for each provider. A model gateway addresses that by placing a single OpenAI-compatible endpoint in front of many providers, so teams keep one key, one base URL, and one invoice.

For models from Chinese labs, the friction is higher: signing up usually requires a Chinese phone number, a domestic payment method such as Alipay or WeChat Pay, and a console only available in Chinese, and AICraft says many teams give up before getting a key. AICraft positions its gateway as removing those steps from the critical path, so the whole catalogue is reachable in English through the same endpoint used for Western models.

Before standardising on a gateway, also check: OpenAI compatibility for existing client libraries and observability; catalogue size and how quickly new models appear; failover behaviour when a provider degrades; consolidated billing instead of per-provider top-ups; and where the gateway and its models are hosted, which determines applicable data-protection rules. AICraft notes the gateway pays for itself when a team needs more than one model family, wants to A/B models without rebuilding integration, or needs a catalogue that is not practically reachable from where the team is.

FAQ

How do I switch between models on the gateway?: Models are addressed as vendor/model, such as deepseek/deepseek-v4-flash or anthropic/claude-opus-4-8, so switching is a string change rather than a migration.
Do I need a Chinese phone number or payment method to reach Chinese models?: No, according to AICraft. You sign up once, in English, and the whole catalogue, including Chinese and Western models, becomes reachable through the same endpoint.
When is a direct provider call simpler than using a gateway?: If you only ever use a single Western model and deploy in one region, calling that provider directly is the simpler answer, and a gateway adds a hop for no benefit.

Zenn AI/MLRead Original Article

Send on LINE

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Get it free

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.