One API for chat, images, video, and more
380+ models from OpenAI, Anthropic, Google and 53 more labs, below market price. Every request goes to the provider that is working best right now.
- Models
- 380+
- Tokens processed
- 376B
- Users
- 35K
import os
from openai import OpenAI
client = OpenAI(
base_url="https://api.naga.ac/v1",
api_key=os.environ["NAGA_API_KEY"],
)
response = client.responses.create(
model="deepseek-v4.1-flash-1017",
input="Hello",
)
print(response.output_text)Models from 56 labs, including
Text, images, video and audio in one API
One key and one balance for every kind of model, through the SDK you already use.
Chat, reasoning and coding models from every major lab, with tools and vision where the model supports them.
Browse in the catalog
DeepSeek V4.1 Flash 1017Deepseek
GLM 5.3 FlashZ.ai
MiniMax M3Minimax
GPT OSS 20BOpenAI
Claude Haiku 5.5Anthropic
Routing that measures every provider
Most models are served by several providers. NagaAI ranks them on live traffic and sends each request to the one that is working best right now.
- Measured on live traffic
- We track how often each provider answers and how fast, for every model, and send each request to the one doing best right now.
- Failures stay on our side
- If a provider errors or stops answering, your request moves to the next one. Providers that are down are skipped before your request reaches them.
- You pay for the answer
- Attempts that failed before another provider answered cost nothing.
Care more about speed than reliability? Pick another routing strategy for a request or for a whole key. Routing strategies
Flagship models, below market price
Pay per token from one balance, with no subscription. Next to our price, struck through, is the lowest price elsewhere.
| Model | Input / 1M | Output / 1M |
|---|---|---|
Claude Opus 5.5Anthropic | $2 | $10 |
GPT-6 AstraOpenAI | $2.50 | $12.50 |
 Gemini 4 ArgonGoogle | $0.50 | $2.50 |
 Grok 4.7xAI | $1 | $3 |
DeepSeek V4 Pro 0423Deepseek | $0.45 | $0.75 |
 Qwen3.8 MaxQwen | $1 | $3 |
 Kimi K3.1 ProMoonShotAI | $1 | $5 |
 Muse WonderMeta Llama | $2 | $10 |
GLM 5.3 PrimeZ.ai | $1.40 | $4.40 |
 Mistral Large 4MistralAI | $0.35 | $1.05 |
Works with the code you already have
Every chat model works with the OpenAI and Anthropic SDKs. Change the base URL and the key, and keep your code, your agent and your prompts.
- OpenAI ResponsesOpenAI SDK, Codex CLI, OpenAI Agents SDK
- OpenAI Chat CompletionsOpenAI SDK and most OpenAI-compatible tools
- Anthropic MessagesAnthropic SDK, Claude Code
Providers differ, your request does not: NagaAI adapts it to each provider, so the same code works with every model. What changes
Start in three steps
- 01Create an account
Sign up with Google, GitHub, Discord or email.
- 02Create an API key
Create a key in the dashboard. It opens with a first request ready to run, and free models answer right away.
- 03Add credits for paid models
Pay by card, PayPal, crypto, Alipay or WeChat Pay. No subscription, and credits do not expire.
From the blog
· Update
Every chat model on Responses, Chat Completions and MessagesOne model, three request formats. Use the Responses API, Chat Completions or Anthropic Messages with any chat model in the catalog.
· News
Video generation in the APIKling, Grok Imagine, Wan and MiniMax video models with the same key and balance, priced per second, with nothing charged for a failed job.
· Feature
Referral program: earn on every top-upEarn 10% of every top-up your referrals make, 15% once 20 of them have paid. They get a 25% bonus on their first top-up.