Skip to content
AkaRouter

One API for 50+ AI models. Pay as you go.

An OpenAI- and Anthropic-compatible gateway for GPT, Claude, Gemini, DeepSeek, GLM, Qwen, Kimi, and Grok. Top up via QRIS, create an API key, and start calling, with real-time cost and latency dashboards.

# Swap the base URL, use an API key from the dashboard
export OPENAI_API_KEY="your-api-key"
curl https://router.aka.ai.id/v1/chat/completions \
-H "Authorization: Bearer $OPENAI_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"claude-sonnet-5-5","messages":[{"role":"user","content":"Hello!"}]}'
AI models
50+
model families
13
API standards supported
2
instant local top-up
QRIS

Everything an AI API should have

All the essentials for running 50+ AI models in production, from the first connection to cost monitoring.

50+ models, one endpoint

GPT, Claude, Gemini, DeepSeek, GLM, Qwen, Kimi, and Grok behind a single base URL. Switching models is a one-string change: no new SDKs, no new vendor accounts.

https://router.aka.ai.id/v1

Instant QRIS top-up

Fund your balance from any QRIS e-wallet or mobile banking app. Credits land automatically. Pay-as-you-go, no subscription, no credit card.

OpenAI & Anthropic compatible
Chat Completions, Responses, Messages, and Embeddings endpoints. Claude Code, Cursor, Cline, LangChain, and Vercel AI SDK work by just changing the base URL.
Real-time usage dashboard
Tokens, cost, TTFT, and speed for every request, with per-key breakdowns so you always know what each project spends.
API keys with full control
Set budget caps, restrict models, whitelist IPs, and apply rate limits per key. Revoke anytime. Keys are stored hashed, so a leak never exposes the secret.
Playground & Compare
Test prompts in the playground, then compare up to 3 models side by side on the same prompt, with cost and speed for each.

From zero to first request in minutes

No sales calls, no infrastructure to manage. Sign up, top up, and route your first request right away.

Start step 1 at router.aka.ai.id (opens in a new tab)
  1. 01Sign up & top up

    Create your account and add balance via QRIS from any e-wallet or mobile banking, no credit card needed.

  2. 02Create an API key

    Set a budget cap, pick allowed models, and add IP whitelists and rate limits per key, from the dashboard.

  3. 03Point your base URL

    Use https://router.aka.ai.id/v1 with any OpenAI SDK, or the Messages endpoint with Anthropic SDKs and Claude Code.

  4. 04Call & monitor

    Route requests to 50+ models and watch tokens, cost, TTFT, and speed update in real time.

Pay-as-you-go, priced per token

No subscription. Top up once with QRIS and spend it across every model. Here are per-1M-token prices for a selection of the most popular models. The full list of 50+ models is always at GET /v1/models.

58 active modelsprices synced

Top up & see every price in the dashboard (opens in a new tab)
ModelFamilyInput / 1MOutput / 1MStatus
claude-opus-5-5Claude$4.80$24.00Active
gpt-5.6-solGPT$4.80$24.00Active
gemini-3.1-proGemini$2.40$14.40Active
claude-sonnet-5-5Claude$2.40$12.00Active
gemini-3.5-flashGemini$1.80$10.80Active
glm-5.3GLM$1.68$5.28Active
deepseek-v4-proDeepSeek$1.58$4.75Active
qwen3.8-maxQwen$1.20$3.60Active
gpt-5.6-lunaGPT$0.24$1.44Active
glm-5.3-flashGLM$0.018$0.30Active

Prices are per 1M tokens (input / output) and may change with provider pricing. The table above shows a curated selection of 10 models; all 50+ models (including Kimi, Grok, MiniMax, and more) with up-to-date prices are available via the GET /v1/models endpoint.

GET /v1/models → live model & price list
  • Pay-as-you-go
  • No subscription
  • QRIS top-up

Frequently asked questions

Short answers about access, payment, models, and privacy.

What is AkaRouter?

An Indonesian AI API platform: one endpoint compatible with both the OpenAI and Anthropic APIs, covering 50+ models from GPT, Claude, Gemini, DeepSeek, GLM, Qwen, Kimi, and Grok. One account, one balance, one base URL.

Which APIs and tools is it compatible with?

OpenAI endpoints (Chat Completions, Responses, Embeddings) and the Anthropic Messages endpoint are available. Claude Code, Cursor, Cline, LangChain, and the Vercel AI SDK work by pointing them at the AkaRouter base URL, with no other code changes.

How does payment work?

Top up via QRIS from any e-wallet or mobile banking app; the balance lands automatically. Usage is billed per token (pay-as-you-go), with no monthly subscription and no credit card.

Which models are available?

50+ models from the GPT, Claude, Gemini, DeepSeek, GLM, Qwen, Kimi, and Grok families. The active list with current input/output prices is available via the GET /v1/models endpoint.

Are my prompts stored?

No. Prompt and response content is never stored, only billing metadata such as token counts and cost. Each deployment also runs on its own isolated instance and database.

How safe is my API key?

API keys are stored as hashes, credentials are encrypted, and admins use 2FA. Every key can be limited by budget, models, IPs, and rate limits, and revoked at any time from the dashboard.

Can I try models before committing?

Yes. Use the Playground to test prompts, or Compare mode to run up to 3 models side by side on the same prompt, with cost and speed shown for each.

What does it cost?

You pay per token, only for what you use. Example per-1M-token prices (input / output): claude-opus-5-5 $4.80 / $24.00, gemini-3.1-pro $2.40 / $14.40, deepseek-v4-pro $1.584 / $4.752, glm-5.3-flash $0.018 / $0.30. The full list is always at GET /v1/models.

Ready to route your first request?

Create an API key, top up with QRIS, and call any of the 50+ models within minutes.