AI PagesModels
EN
Storefront
C
Quick StartChatUsageModelsAPI Keys
1 API Key • All AI Models • Prepaid IDR Billing

Choose Any Model. Zero Subscription Hassle.

With your single sk-gw-live-... key, you can dynamically query Claude, DeepSeek, GPT-4o, and Gemini. Prepaid token credits are consumed in real-time in Indonesian Rupiah (IDR) per 1M tokens with automatic prompt cache discounts.

View Quick Start CodeTest in AI Chat

Prepaid Token Consumption Calculator

Estimate exact IDR deductions based on Input, Cache Hit, and Output tokens.

1k250k500k
0% (Cold)50%100% (Warm)
50016k32k
Estimated API Call Deduction:
15,000 fresh input + 35,000 cached + 10,000 completion
Rp 560
~$0.0350 USD
Showing 16 models

Claude 3.7 Sonnet (Hybrid Reasoning)

Latest Frontier
anthropic/claude-3-7-sonnet
Active

First hybrid model with dynamic reasoning budget. Exceptional for complex architecture and programming.

Extended ThinkingPrompt CachingCode ExpertVision
Input / 1MRp 48.000
Cache Read-90%
Rp 4.800
Output / 1MRp 240.000
Context: 200k tokensTest Model

Claude 3.5 Sonnet

Best for Coding
anthropic/claude-3-5-sonnet
Active

Industry benchmark for software engineering, complex system design, and multimodal agent execution.

Prompt CachingTool CallingArtifactsVision
Input / 1MRp 48.000
Cache Read-90%
Rp 4.800
Output / 1MRp 240.000
Context: 200k tokensTest Model

Claude 3.5 Haiku

Sub-Second Speed
anthropic/claude-3-5-haiku
Active

Ultra-fast sub-second response times with Sonnet-grade code generation for high-throughput microservices.

Ultra-FastPrompt CachingTool Calling
Input / 1MRp 12.800
Cache Read-90%
Rp 1.280
Output / 1MRp 64.000
Context: 200k tokensTest Model

DeepSeek R1

Top Price/Performance
deepseek/deepseek-r1
Active

Open-weights reasoning powerhouse rivaling OpenAI o1 at 90% lower token cost. Full chain-of-thought support.

Reasoning TracesMath & AlgorithmsPrompt CachingJSON Mode
Input / 1MRp 8.800
Cache Read-75%
Rp 2.240
Output / 1MRp 35.000
Context: 128k tokensTest Model

DeepSeek V3 (MoE)

deepseek/deepseek-chat
Active

671B parameter Mixture-of-Experts foundation model delivering near-frontier performance at minimal token cost.

MoE ArchitecturePrompt CachingHigh Concurrency
Input / 1MRp 4.320
Cache Read-75%
Rp 1.120
Output / 1MRp 17.600
Context: 128k tokensTest Model

GPT-4o Omnimodel

Flagship Multimodal
openai/gpt-4o
Active

OpenAI flagship multimodal model with native image understanding, audio comprehension, and structured outputs.

VisionAudioStructured OutputsFunction Calling
Input / 1MRp 40.000
Cache Read-50%
Rp 20.000
Output / 1MRp 160.000
Context: 128k tokensTest Model

GPT-4o mini

openai/gpt-4o-mini
Active

Extremely affordable omnimodel designed for lightweight tasks, summaries, and high-frequency background jobs.

VisionFast InferenceStructured Outputs
Input / 1MRp 2.400
Cache Read-50%
Rp 1.200
Output / 1MRp 9.600
Context: 128k tokensTest Model

OpenAI o3-mini

New Reasoning
openai/o3-mini
Active

Latest-generation reasoning model optimized for STEM, competitive programming, and complex multi-file logic.

STEM ReasoningAdjustable EffortTool Calling
Input / 1MRp 17.600
Cache Read-50%
Rp 8.800
Output / 1MRp 70.400
Context: 200k tokensTest Model

OpenAI o1

openai/o1
Active

High-compute reasoning model designed to spend more time thinking before answering hard scientific problems.

Deep ReasoningVision InputMath & Code
Input / 1MRp 240.000
Cache Read-50%
Rp 120.000
Output / 1MRp 960.000
Context: 200k tokensTest Model

Gemini 2.0 Flash

1M Context
google/gemini-2.0-flash
Active

Next-gen Google multimodal foundation model with 1M context window and near-instant inference speed.

1M ContextMultimodalSpeed LeaderAudio/Video
Input / 1MRp 1.600
Cache Read-75%
Rp 400
Output / 1MRp 6.400
Context: 1,000,000 tokensTest Model

Gemini 2.0 Flash Thinking

google/gemini-2.0-flash-thinking
Active

Combines the speed of Flash with visible reasoning chains and 1M context window processing.

Reasoning Traces1M ContextMultimodal
Input / 1MRp 2.400
Cache Read-75%
Rp 600
Output / 1MRp 9.600
Context: 1,000,000 tokensTest Model

Gemini 1.5 Pro

2M Context Monster
google/gemini-1.5-pro
Active

Massive 2 Million token context window. Ingest whole codebases, hours of video, or entire textbooks in one call.

2M ContextFull Codebase IngestionMultimodal
Input / 1MRp 20.000
Cache Read-75%
Rp 5.000
Output / 1MRp 80.000
Context: 2,000,000 tokensTest Model

Meta Llama 3.3 (70B)

Open Weights Leader
meta-llama/llama-3.3-70b-instruct
Active

State-of-the-art open source instruction model rivaling proprietary 400B models at a fraction of compute cost.

Open WeightsTool CallingHigh Throughput
Input / 1MRp 6.400
Cache Read-75%
Rp 1.600
Output / 1MRp 12.800
Context: 128k tokensTest Model

Meta Llama 3.1 (405B)

meta-llama/llama-3.1-405b-instruct
Active

The largest and most capable open-weights foundation model in existence. Highly suitable for synthetic data & distillation.

405B ParametersFrontier Open WeightsMultilingual
Input / 1MRp 32.000
Cache Read-75%
Rp 8.000
Output / 1MRp 64.000
Context: 128k tokensTest Model

Qwen 2.5 Coder (32B)

Code Specialist
qwen/qwen-2.5-coder-32b-instruct
Active

Top-tier code generation model matching GPT-4o on Python, TypeScript, and SQL code completion tasks.

Code CompletionFill-in-the-MiddleAgent Ready
Input / 1MRp 4.800
Cache Read-73%
Rp 1.280
Output / 1MRp 14.400
Context: 128k tokensTest Model

Mistral Large 2

mistralai/mistral-large-2
Active

Flagship multilingual reasoning and function calling model from Mistral AI with 123B parameters.

MultilingualFunction CallingStrict JSON
Input / 1MRp 32.000
Cache Read-75%
Rp 8.000
Output / 1MRp 96.000
Context: 128k tokensTest Model

How Single-Key Routing & IDR Prepaid Consumption Works

1

Single API Secret Key

Generate one key from the API Keys tab. You don't need Anthropic, OpenAI, or Google accounts.

2

Switch by Model ID

Just pass any model string (e.g. anthropic/claude-3-7-sonnet or deepseek/deepseek-r1) in standard OpenAI requests.

3

Prepaid IDR Balance

Top up prepaid credits via QRIS or BCA/Mandiri VA. Tokens are deducted accurately based on Input, Cache Read, and Output rates.