# JigsawStack Pricing Updates & v3 Focus!

JigsawStack began with a simple goal: **make infrastructure automation accessible to everyone**. Along the way, we connected deeply with developers and discovered the critical aspects of infrastructure that truly matter.

> Full video walkthrough with the JigsawStack team [here](https://www.youtube.com/watch?v=NxNdhmV0KS0&t=49s).

## If you’re not gonna read everything, read this 👇

* **Rollout Date:** April 11th  
    *(No action required on your part; the switch to the new pricing is automatic.)*
    
* **Notifications:**  
    Two additional emails and a dashboard notice will be sent as we approach the update.
    
* **Enterprise Customers:**  
    Custom pricing for the commitment tier remains unchanged until the next contract cycle (unless requested otherwise).
    
* **v3 updates**
    
    Progressive weekly updates from April 11th
    

## From v1 to v2: Delivering Real Value

When we launched **JigsawStack v2**, we adopted a clear mantra:

> *If a feature only adds convenience, we won't build it.*

Instead, we focused on delivering features that save developers months of work. This meant streamlining our services—removing those that didn’t provide tangible value and honing in on the essentials.

### Our Three Key Model Categories

* **Data Extraction**
    
    * AI Scraper
        
    * vOCR
        
    * Speech-to-Text
        
* **Data Transformation**
    
    * Translation
        
    * Prompt Engine
        
    * Text-to-Speech
        
* **Data Validation**
    
    * NSFW Check
        
    * Spell Check
        

Our approach is simple yet powerful:

![](https://cdn.hashnode.com/res/hashnode/image/upload/v1743558667970/9864c575-fb9d-4cd1-bdad-1ebba0df0cb1.png align="center")

> **Base Infra + SLM/Deep Learning Model (trained/fine-tuned for base infra) + ML Model (grounding) = JigsawStack**

Training our models to manage underlying infrastructure opened up new possibilities, reduced limitations, and increased accuracy.

## Introducing JigsawStack v3

JigsawStack v3 builds on our commitment to real value by:

* **Merging and removing services** to eliminate redundancy.
    
* **Improving accessibility** with major updates.
    
* **Lowering costs and scaling solutions** for developers at every stage.
    

We emphasize **data** because infrastructure is fundamentally about data movement, driving our commitment to our three key categories.

### Upcoming Service Updates

#### Service Merges

* **AI Scraper & Web Scraper** combined
    
* **AI Web Search & Geo Search** combined
    

**Deprecation**

* Web Scraper (Merged with AI Scraper)
    
* Geo Search (Merged with Web Search)
    

#### New Model Launches

* **Embedding model** (v1 beta)
    
* **Image-to-Image Translation model**
    
    * V1: In-scene machine translation
        
    * V2: Diffusion based
        
* **Speaker Diarization model** (faster, cheaper, and more accurate)
    
* **SOTA Object detection** by combing SAM2 + Object Labeling + OmniParserv2 + vLLM
    

#### Enhanced Features

* Updated image generation service with native support for:
    
    * Image-to-image
        
    * Text-to-image
        
    * Image-to-video
        
    * Prompt-based editing
        
* Improved Python SDK (upcoming breaking change)
    
* Usable apps built on JigsawStack models
    

## Pricing Updates: Scalable and Developer-Friendly

Accessibility means having **predictable and scalable pricing**. In v1, a fixed usage-based pricing model on API calls worked well. However, with our enhanced models in v2, usage is truly driven by your input.

Now, we’ve moved to a **token-based pricing model** that is granular enough for developers to control their costs while scaling dynamically with usage. This model aligns with market standards for many AI companies.

### New Pricing Tiers

| Tier | Price | Tokens Included | Additional Cost |
| --- | --- | --- | --- |
| Free tier | $0/month | 1 million tokens | \- |
| Pro tier | $27/month | 8 million tokens | $1.40 per million tokens |

Tokens are calculated across three dimensions:

* **Input & Output Tokens:** $1.40/million tokens
    
* **Inference Time Tokens:** 1 millisecond (ms) processing = 1 token
    

> *A consistent rate of $1.40 per million tokens applies across all models*

### How Token-Based Pricing Works

Every request now includes a `_usage` key providing the token breakdown, so you can better estimate costs based on:

* **Input tokens**
    
* **Output tokens**
    
* **Inference time tokens**
    

```json
{
  "_usage": {
    "input_tokens": 43,
    "output_tokens": 9470,
    "inference_time_tokens": 16569,
    "total_tokens": 26082
  }
}
```

#### Example 1

* **Input:** 100 tokens
    
* **Processing:** 3 seconds (3000 ms)
    
* **Output:** 20 tokens
    

Calculation:

```plaintext
Cost per token (CPT) = 1.40 / 1,000,000
Total cost ≈ 100 * CPT + 20 * CPT + 3000 * CPT ≈ $0.004368
```

#### Example 2

* **Input:** 3 tokens (URL only)
    
* **Output:** 100 tokens
    
* **Processing:** 5 seconds (5000 ms)
    

Calculation:

```plaintext
Total cost ≈ 3 * CPT + 100 * CPT + 5000 * CPT ≈ $0.0071442
```

In these examples, processing time is factored into the cost—not the duration of the audio or the input size—resulting in savings of **6x to 10x** compared to the previous flat rate of $0.05 per request.

## How Does This Compare?

### Token-Based Pricing ($1.40/1m Tokens) vs. $0.05 per Invocation

* **Prediction API:**
    
    ```plaintext
    Token Cost = (1.40 / 1,000,000) * 984 ≈ 0.0013776  
    Savings Ratio ≈ 0.05 / 0.0013776 ≈ 36x cheaper
    ```
    
* **AI Web Scraper API:**
    
    ```plaintext
    Token Cost = (1.40 / 1,000,000) * 13,507 ≈ 0.0189098  
    Savings Ratio ≈ 2.65x cheaper
    ```
    
* **vOCR API:**
    
    ```plaintext
    Token Cost = (1.40 / 1,000,000) * 15,068 ≈ 0.0210952  
    Savings Ratio ≈ 2.37x cheaper
    ```
    
* **NSWF API:**
    
    ```plaintext
    Token Cost = (1.40 / 1,000,000) * 1,524 ≈ 0.0021336  
    Savings Ratio ≈ 23.44x cheaper
    ```
    
* **Image Generation API:**
    
    ```plaintext
    Token Cost = (1.40 / 1,000,000) * 13,297 ≈ 0.0186158  
    Savings Ratio ≈ 2.69x cheaper
    ```
    

## Rollout and Transition for Existing Users

* **Rollout Date:** April 11th  
    *(No action required on your part; the switch to the new pricing is automatic.)*
    
* **Notifications:**  
    Two additional emails and a dashboard notice will be sent as we approach the update.
    
* **Enterprise Customers:**  
    Custom pricing for the commitment tier remains unchanged until the next contract cycle (unless requested otherwise).
    
* **v3 updates**
    
    Progressive weekly updates from April 11th
    

## We Value Your Feedback

JigsawStack v3 is built with a developer-first mindset—improving everything from developer experience (DX) to cost efficiency. Your honest and direct feedback helped shape v3, and we’re eager to hear your thoughts on how we can continue to improve.

Our small, passionate team of engineers is growing, and we’re always looking for talented, “cracked” engineers to join us. If you're interested, check out our [careers page](https://jigsawstack.com/careers) or reach out—even if no current role fits your profile, you might just be the perfect addition to our team!

Thank you for being part of our journey! If you encounter any issues during this transition or notice any bugs, please email us at **support@jigsawstack.com** or drop a message on our [Discord](https://discord.gg/dj8fMBpnqd). We're here to help.