# Synthetic

> AI platform providing access to multiple LLMs with subscription or usage-based pricing, offering both UI and API access.

Synthetic is an AI platform that provides unified access to a wide range of large language models (LLMs) through both a user-friendly interface and API. The platform offers flexible pricing options including subscription plans and usage-based billing, making it accessible for individuals, professionals, and enterprise users. With support for always-on models and on-demand GPU computing, Synthetic enables users to leverage cutting-edge AI models without complex infrastructure setup.

**Key Features:**

- **Multi-Model Access** - Provides access to numerous always-on models including DeepSeek, Meta Llama, MiniMax, Kimi, Qwen, and OpenAI models, all included in subscription plans with no per-token billing.

- **Dual Access Methods** - Offers both a friendly chat UI and full API access, allowing users to interact with models through their preferred interface.

- **Competitive Rate Limits** - Standard plan offers 3x higher rate limits than Claude's $20/month plan, while Pro plan provides 10x higher limits with 135 and 1,350 requests per five hours respectively.

- **LoRA Support** - Supports Low-rank adapters (LoRAs) for fine-tuned models, enabling specialized task performance on top of base models like Llama 3.2 and Meta-Llama 3.1.

- **On-Demand GPU Computing** - Allows launching any Hugging Face model on-demand with automatic GPU detection and competitive per-minute pricing starting at 1.2 cents/min per GPU.

- **Extended Context Windows** - Supports models with context lengths ranging from 128k to 256k tokens, enabling processing of large documents and conversations.

- **Smart Request Counting** - Small requests with less than 2048 input/output tokens count as only 0.2 requests, and tool call messages count for even less at 0.1 requests.

- **Flexible Pricing** - Choose between flat monthly subscriptions for predictable costs or pay-per-use for enterprise needs and custom models.

To get started, users can sign up on the platform and choose between the Standard subscription at $20/month for individual use, Pro at $60/month for heavy users, or usage-based pricing for enterprise requirements. The platform automatically handles model deployment and GPU allocation, making it straightforward to access powerful AI capabilities.

## Features
- Access to multiple always-on LLMs
- UI and API access
- LoRA model support
- On-demand GPU computing
- Automatic GPU detection and allocation
- Extended context windows up to 256k tokens
- Smart request counting for subscriptions
- Pay-per-token for always-on models (usage-based)
- Pay-per-minute for on-demand models
- Hugging Face model integration

## Integrations
Hugging Face, DeepSeek, Meta Llama, MiniMax, Moonshot Kimi, Qwen, OpenAI GPT-OSS, GLM

## Platforms
WEB, API

## Pricing
Subscription-based

## Links
- Website: https://synthetic.new
- Documentation: https://dev.synthetic.new/docs/api/overview
- EveryDev.ai: https://www.everydev.ai/tools/synthetic