One key works across every tool. Pick yours below. The API is live and self-serve — sign up, create a key, and you're running in minutes.
export ANTHROPIC_BASE_URL="https://api.gloritoken.ai" export ANTHROPIC_AUTH_TOKEN="sk-your-key" export ANTHROPIC_MODEL="qwen3.8-max"
Base URL: https://api.gloritoken.ai/v1 API Key: sk-your-key Model ID: qwen3.8-max
OpenAI Base URL: https://api.gloritoken.ai/v1 OpenAI API Key: sk-your-key Custom model: qwen3.8-max
from openai import OpenAI client = OpenAI( base_url="https://api.gloritoken.ai/v1", api_key="sk-your-key", ) resp = client.chat.completions.create( model="qwen3.8-max", messages=[{"role":"user","content":"Write an LRU cache"}], )
One key, pick per request by model name. Every listed model is the maker's official API, priced up to 48% below the vendors' international rates, cached input discounted at the same ratio.
| Model | Type | Notes | In /1M | Out /1M |
|---|---|---|---|---|
| qwen3.8-max | Flagship | Alibaba's 2.4T MoE flagship, 1M ctx, native vision | $1.04 | $3.11 |
| kimi-k3 | Flagship | Moonshot 2.8T, 1M ctx, native vision — price cut Aug 2026 | $1.73 | $8.65 |
| glm-5.2 | Flagship | Zhipu's newest, coding + general, 1M ctx | $0.74 | $2.57 |
| qwen3.7-plus | Flagship | Multimodal, best value, 256K ctx | $0.18 | $0.74 |
| deepseek-v4-pro | Flagship | All-round reasoning flagship, 1M ctx — vendor parity | $0.435 | $0.87 |
| deepseek-v4-flash | Value | High-volume & batch workhorse, 1M ctx | $0.09 | $0.18 |
| qwen3.7-flash | Value | Vision-language Flash, multimodal agents, 1M ctx | $0.02 | $0.07 |
| minimax-m3 | Flagship | Coding & agentic, native multimodal, 1M ctx | $0.39 | $1.54 |
| kimi-k2.7-code | Coding | Moonshot dedicated coder, 256K ctx | $0.60 | $2.48 |
| kimi-k2.7-code-highspeed | Exclusive | ~180–260 tok/s, no international provider offers it | $1.89 | $7.85 |
| glm-5.2-fast-preview | Exclusive | Speed-tuned GLM 5.2, 1.5–2× output throughput, 1M ctx | $1.47 | $5.15 |
| qwen3.7-max | Legacy | Previous-gen flagship for pinned workflows, 1M ctx | $1.10 | $3.31 |
* Prices are flat per 1M tokens as billed by our gateway. All listed models are live now — official, full-quality builds from the makers' own APIs.
Your requests are processed on our Singapore gateway; inference runs on the model makers' official APIs, served from mainland China — stated plainly. We don't store prompt contents and never train on your data; we keep only usage metadata for billing. We do not relay Claude, GPT or Gemini.
No. We keep only the usage metadata needed for billing (token counts, model, timestamps). Prompt and completion contents are not stored and never used for training.
Yes. We expose a real Anthropic-protocol endpoint (/v1/messages), so Claude Code connects with two environment variables — no proxy hacks, no wrapper. Streaming, tool use and long context all work.
Signing up gets you free trial credit — roughly 5.5M output tokens on deepseek-v4-flash, enough for days of real work. To add more, top up by card ($10, $20 or $50) through Stripe's secure checkout — use your account email at checkout and we'll credit your balance the same day, usually within minutes; your API key never changes. Prefer an invoice or a custom amount? Email hello@gloritoken.ai or message us on WhatsApp. Credit never expires and your balance stops at zero, so you can never be overbilled. Consumed usage is non-refundable; any unused balance can be refunded on request within 30 days.
No, and we never will — we are Chinese-frontier-model specialists, first-hand supply only. No gray-market relays, no "mirrors". That focus is exactly why you can trust the models you get are what they claim to be.