Kimi K3 + GitHub OAuth + Smart Usage Dashboard — now live on AIBridge

#ai #api #llm #kimi

I shipped three things that actually matter for an API product.

1. Kimi K3 — reasoning without a toggle

Most reasoning models make you decide: think mode ON or OFF. Kimi K3 doesn't ask. It always reasons. Every request runs through a thinking pass before generating the answer.

1M context window. 4x what Claude gives you. 8x what o1 gives you. Drop a codebase, get a code review. No chunking required.

import openai
client = openai.OpenAI(
api_key="mb-xxx",
base_url="https://aibridge-api.com/v1"
)
client.chat.completions.create(
model="kimi-k3",
messages=[{"role":"user","content":"Review this entire repo for bugs"}]
)

One of 15 Chinese AI models. Same endpoint for all of them.

2. GitHub OAuth — signup in 5 seconds

Old flow: email → password → verify inbox → type 6-digit code → get API key.

New flow: click GitHub → authorize → dashboard.

Five seconds. No email. No code. No friction.

Registration page and login page both have it. One click, you're in.

3. Smart usage dashboard

The dashboard now tells you exactly where you stand:

Usage: 385,000 / 500,000 tokens [████████░░] 77%

💡 77% used. Upgrade to Pro for 10x more → $9.90/mo

The bar escalates as you approach the limit — blue below 50%, orange at 80%, red at 90% with a flashing warning. The upgrade button goes from subtle to "you really need to see this" at the exact moment it matters.

Also included

Free Playground — try all 15 models with no signup. 10 requests/day. aibridge-api.com/playground.html
Anti-bot protection — verification code endpoint rate-limited to 1 request/minute per IP
Auto-healing status — model health dashboard resets stale errors after 5 minutes

The stack

15 models. One OpenAI-compatible endpoint. One API key.

DeepSeek — V4 Pro, V4 Flash, Chat, Reasoner, Coder
Qwen — Qwen3-235B, Max, Plus
GLM — GLM-4 Plus, Air, Flash
Moonshot — Kimi K3, Kimi 128K, Kimi 32K, Kimi 8K