BrushLLM [ 全球入口 ]
[ 中国大模型全球枢纽 ] / [ OPENAI COMPATIBLE ] / [ GLOBAL ROUTING ]

Chinese Large Language Models, Accessible Worldwide

Unleash the cognitive depth of DeepSeek, Qwen 2.5, GLM-4, Moonshot (Kimi), and MiniMax. One unified API gateway with zero payment hurdles and low global latency.

100%
OpenAI Compatible
Drop-in replacement for standard SDKs and tools.
< 200ms
Global Edge Relay
Optimized cross-border routing via Cloudflare network.
5x - 10x
Cost Efficiency
Fraction of western frontier model pricing.
Zero Friction
No Mainland Phone
No Alipay/WeChat or Chinese KYC needed.
[ 精选模型 ] Curated Frontier Models

Frontier Chinese LLMs at Your Fingertips

View all models (6)
[ RECOMMENDED ]
DeepSeek

DeepSeek-V3

[ GENERAL ]

State-of-the-art 671B MoE model (37B active) delivering frontier performance rivaling Claude 3.5 Sonnet and GPT-4o at a tiny fraction of the cost.

Context Window 64k tokens
Pricing (In/Out 1M) $0.14 / 1M / $0.28 / 1M
671B MoE Multi-Head Latent Attention FP8 Native Ultra Low Cost
ID: deepseek-v3
Docs
[ RECOMMENDED ]
DeepSeek

DeepSeek-R1

[ REASONING ]

Groundbreaking open reasoning model utilizing large-scale reinforcement learning. Displays human-like chain-of-thought, self-verification, and mathematical proofing.

Context Window 64k tokens
Pricing (In/Out 1M) $0.55 / 1M / $2.19 / 1M
RL Reasoning AIME 2024 Champion CoT Verification Coding Logic
ID: deepseek-r1
Docs
[ RECOMMENDED ]
Alibaba Cloud

Qwen 2.5 72B

[ CODE ]

Alibaba's flagship open-weights powerhouse. Pretrained on 18T tokens, offering extraordinary coding capability, multi-lingual fluency across 29+ languages, and structured JSON generation.

Context Window 128k tokens
Pricing (In/Out 1M) $0.20 / 1M / $0.60 / 1M
18T Pretraining 29+ Languages 128K Context Superior Coding
ID: qwen-2.5-72b
Docs
[ RECOMMENDED ]
MiniMax

MiniMax Text-01

[ LONG CONTEXT ]

Next-generation 456B MoE architecture with unprecedented 4M token context support. Blends deep world knowledge, nuanced emotional tone, and lightning-fast long-context retrieval.

Context Window 4M tokens
Pricing (In/Out 1M) $0.28 / 1M / $1.12 / 1M
4M Context Window 456B MoE Linear Attention Hybrid Cost Leader
ID: minimax-text-01
Docs
[ 极速接入 ] Sub2API Gateway

One Endpoint.
Every Chinese Frontier Model.

No need to manage multiple API accounts, separate payment currencies, or navigate complex Chinese cloud consoles. BrushLLM's Sub2API Gateway provides a unified OpenAI-compatible endpoint with instant token provisioning.

  • Drop-in Replacement: Compatible with LangChain, LlamaIndex, LiteLLM, and AutoGen.
  • Sub-200ms Latency: Global Cloudflare Anycast edge acceleration.
  • International Payments: Pay via Stripe, Credit Card, or Crypto (USDT).
# 100% OpenAI SDK Compatible
from openai import OpenAI

client = OpenAI(
    base_url="https://api.brushllm.com/v1",
    api_key="sk-brush-your-token-here"
)

response = client.chat.completions.create(
    model="deepseek-v3",  # Or 'qwen-2.5-72b', 'glm-4-plus'
    messages=[
        {"role": "user", "content": "Summarize the latest breakthrough in Chinese open weights LLMs."}
    ],
    stream=True
)

for chunk in response:
    print(chunk.choices[0].delta.content or "", end="")
[ 无界互联 ]

Why Global Developers Choose BrushLLM

Accessing Chinese frontier AI from outside mainland China traditionally poses significant friction. BrushLLM removes every barrier.

Direct Mainland Access

[ Friction ]
  • × Requires mainland Chinese (+86) mobile phone for SMS verification.
  • × Only accepts Chinese bank cards, WeChat Pay, or Alipay.
  • × Cross-border network packet loss and high latency (600ms+).
  • × Fragmented API schemas across disparate cloud vendor dashboards.
[ THE BRUSHLLM WAY ]

BrushLLM Global Gateway

[ Seamless ]
  • Instant Sign-Up: Register with Google or GitHub, zero KYC.
  • Global Payment: Pay with international credit cards or crypto.
  • Cloudflare Edge: Sub-200ms round-trip latency globally.
  • 100% OpenAI Format: Change only `base_url` and `api_key`.
[ 即刻启程 ]

Experience the Frontier of Chinese AI Today

Start building intelligent applications with DeepSeek, Qwen 2.5, and GLM-4. Zero lock-in, pay as you go.