
NVIDIA is [reportedly](https://www.reuters.com/world/china/nvidia-working-new-ai-chip-china-that-outperforms-h20-sources-say-2025-08-19/) working on a new China-specific AI chip that will be more powerful than the aging HGX H20 model the company is currently authorized to sell in China.

Recent reports [suggest](https://www.tomshardware.com/tech-industry/artificial-intelligence/nvidia-responds-to-reports-that-its-h20-gpu-for-china-is-ending-production-next-gen-b30a-green-light-up-to-the-united-states-government-according-to-ceo-jensen-huang) that the H20 – originally developed as a scaled-down alternative to the A100 and H100 that were banned from export to China – is nearing the end of its production run. Its successor: the tentatively named B30A chip.

Based on the latest Blackwell architecture, the B30A will feature a single-die design delivering roughly half the raw computing power of NVIDIA's flagship B300 accelerator, according to sources. The new chip will include high-bandwidth memory and NVLink support for fast data transfer between processors – both of which are also present in the Hopper-based H20.

NVIDIA aims to send test batches of the B30A chip to Chinese customers as early as sometime this month, sources said.

The move comes following remarks from President Donald Trump suggesting that more advanced NVIDIA chips could be cleared for sale in China. However, regulatory approval remains uncertain. There are ongoing concerns about granting China broader access to U.S. AI technology – which could weaken the United States' competitive edge in the sector.

At the same time, analysts warn that restricting China too heavily risks driving developers toward rivals like Huawei. NVIDIA argues that it's important to retain market share in China – which generated 13% of the company's revenue in the past year – so that its software ecosystem stays relevant in one of the world's largest AI markets.

NVIDIA is also preparing a second Blackwell-based China-specific chip called the RTX6000D, designed primarily for AI inference tasks, which is expected to start shipping this month. With weaker specs than the H20 and bandwidth just under U.S. export limits, it's set to be a compliant, lower-cost alternative within NVIDIA's lineup.

For those outside China, however, these region-specific models highlight the stark difference between scaled-down chips and the unrestricted performance available elsewhere.

### On-Demand Access to Powerful GPUs, Anytime

While NVIDIA adapts its offerings to meet specific market regulations, users elsewhere don't have to work around restricted versions.

On [**Vast.ai**](http://Vast.ai), you can access a wide range of powerful GPUs – from the [RTX 5090](https://vast.ai/article/nvidia-geforce-rtx-5090-specs-everything-you-need-to-know) to the [A100](https://vast.ai/article/nvidia-rtx-4090-vs-a100-two-powerhouses-two-purposes), [H200](https://vast.ai/article/nvidia-h100-vs-h200-two-hopper-based-heavyweights), [B200](https://vast.ai/article/nvidia-h200-vs-b200-comparing-datacenter-grade-accelerators), and more – on demand, at a fraction of the cost of traditional cloud providers. Our cloud GPU rental platform makes it easy. Just choose the hardware you need, spin up an instance in seconds, and pay only for the compute you use.

Get started with Vast today, and enjoy flexible, cost-effective GPU compute that scales with your needs\!


NVIDIA is developing the B30A AI chip for China, a Blackwell-based successor to the H20 with half the power of the B300. Learn what it means for AI exports, competition, and how you can still access unrestricted GPUs like the H200, B200, and A100 on Vast.ai.

Report: NVIDIA Developing New Blackwell-Based AI Chip for China


## Introduction

Qwen-Image revolutionizes image generation with its exceptional ability to render complex text directly within images, supporting multiple languages including English and Chinese with perfect typographic detail. This powerful foundation model from Alibaba's Qwen series goes beyond simple image creation, offering advanced editing capabilities, style transfer, and precise control over visual elements.

What sets Qwen-Image apart is its nuanced understanding of text-image relationships. While other models struggle with text rendering, Qwen-Image seamlessly integrates readable text into visuals, making it ideal for creating signage, posters, infographics, and artistic compositions that require precise text placement and formatting.

Vast.ai provides the perfect platform for deploying Qwen-Image, offering access to high-performance GPUs at competitive prices with the flexibility to scale based on your needs.

In this guide, you'll deploy Qwen-Image on Vast.ai and explore its capabilities for text-integrated image generation.

## Setting Up the Environment

First, install the Vast CLI and configure your API key:

```bash
pip install --upgrade vastai
export VAST_API_KEY="<your-api-key>"
vastai set api-key $VAST_API_KEY
```

You can obtain your API key from the [Vast.ai Account Page](https://cloud.vast.ai/account/).

## Choosing the Right Hardware

Qwen-Image requires substantial GPU resources for optimal performance:

-   **GPU Memory**: Minimum 80GB VRAM for the full model
-   **Recommended Hardware**: NVIDIA A100 80GB or H100
-   **Storage**: 200GB disk space for model weights and dependencies

Search for suitable instances with these specifications:

```bash
vastai search offers "compute_cap >= 750 \
gpu_ram >= 80 \
num_gpus = 1 \
static_ip = true \
direct_port_count >= 1 \
verified = true \
disk_space >= 150 \
rentable = true"
```

This search filters for:

-   Modern GPUs with compute capability 7.5+
-   At least 80GB VRAM for the model
-   Static IP for stable connections
-   Verified providers for reliability
-   Adequate storage for model and outputs

## Deploying Qwen-Image

Deploy the instance using a pre-configured Docker image optimized for transformer models:

```bash
export INSTANCE_ID=<instance-id>
vastai create instance $INSTANCE_ID \
    --image huggingface/transformers-pytorch-gpu \
    --env '-p 8888:8888' \
    --disk 200 \
    --args bash -c "pip install jupyter && jupyter notebook --ip=0.0.0.0 --port=8888 --no-browser --allow-root --NotebookApp.token='' --NotebookApp.password=''"
```

**Key parameters explained:**

-   `--image`: HuggingFace's PyTorch GPU image with ML libraries pre-installed
-   `--env '-p 8888:8888'`: Maps Jupyter's default port for external access
-   `--disk 200`: Allocates 200GB for model weights and generated images
-   `--args`: Auto-installs Jupyter and launches a passwordless notebook server

## Connecting to Your Instance

After deployment:

1. Navigate to the [Instances tab](https://cloud.vast.ai/instances/)
2. Find your instance and wait for "Running" status
3. Look for the "Open Ports" section in instance details
4. Click the port mapping link (e.g., `123.123.123.123:12345 -> 8888/tcp`)
5. This opens Jupyter directly in your browser
6. Create a new notebook or upload the example notebook if available

## Installing Dependencies

Install the required packages in your notebook:

```bash
apt install -y git
pip install git+https://github.com/huggingface/diffusers
```

The latest diffusers library from GitHub includes Qwen-Image support.

## Loading the Model

Initialize Qwen-Image with optimal settings for GPU deployment:

```python
from diffusers import DiffusionPipeline
import torch

model_name = "Qwen/Qwen-Image"

# Load the pipeline
if torch.cuda.is_available():
    torch_dtype = torch.bfloat16
    device = "cuda"
else:
    torch_dtype = torch.float32
    device = "cpu"

pipe = DiffusionPipeline.from_pretrained(model_name, torch_dtype=torch_dtype)
pipe = pipe.to(device)
```

The model uses bfloat16 precision on GPU for optimal memory usage and performance.

## Using Qwen-Image

Now let's explore Qwen-Image's text rendering capabilities with examples:

### Example 1: Complex Text Integration with Multiple Languages

```python
import os

positive_magic = {
    "en": "Ultra HD, 4K, cinematic composition.", # for english prompt
}

# Create img folder if it doesn't exist
os.makedirs("img", exist_ok=True)

# Generate image
prompt = '''A coffee shop entrance features a chalkboard sign reading "Qwen Coffee 😊 $2 per cup," with a neon light beside it displaying "Welcome". Next to it hangs a poster showing a beautiful Chinese woman, and beneath the poster is written "π≈3.1415926-53589793-23846264-33832795-02384197". Ultra HD, 4K, cinematic composition'''
negative_prompt = " " # using an empty string if you do not have specific concept to remove

# Set image dimensions (16:9 aspect ratio)
width, height = 1664, 928

image = pipe(
    prompt=prompt + positive_magic["en"],
    negative_prompt=negative_prompt,
    width=width,
    height=height,
    num_inference_steps=50,
    true_cfg_scale=4.0,
    generator=torch.Generator(device="cuda").manual_seed(42)
).images[0]

# Save image to img folder
image.save("img/example_1.png")

# Display the image
display(image)
```

### Output

![Qwen Coffee Shop with Text](/uploads/example_1.webp)

This example demonstrates Qwen-Image's ability to:

-   Render multiple text elements (chalkboard, neon sign, poster)
-   Handle emojis and special characters
-   Integrate mathematical constants with precise formatting
-   Maintain readability across different text styles

### Example 2: Creative Fantasy Scene

```python
positive_magic = {
    "en": "Ultra HD, 4K, cinematic composition.", # for english prompt
}

# Create img folder if it doesn't exist
os.makedirs("img", exist_ok=True)

# Generate image
prompt = '''Rare photo: UFO hovering over a yeti riding the Loch Ness monster.'''
negative_prompt = " " # using an empty string if you do not have specific concept to remove

# Set image dimensions (16:9 aspect ratio)
width, height = 1664, 928

image = pipe(
    prompt=prompt + positive_magic["en"],
    negative_prompt=negative_prompt,
    width=width,
    height=height,
    num_inference_steps=50,
    true_cfg_scale=4.0,
    generator=torch.Generator(device="cuda").manual_seed(42)
).images[0]

# Save image to img folder
image.save("img/example_2.png")

# Display the image
display(image)
```

### Output

![Fantasy Scene with UFO and Mythical Creatures](/uploads/example_2.webp)

This showcases Qwen-Image's creative capabilities in composing imaginative scenarios.

### Example 3: Futuristic Space Scene

```python
positive_magic = {
    "en": "Ultra HD, 4K, cinematic composition.", # for english prompt
}

# Create img folder if it doesn't exist
os.makedirs("img", exist_ok=True)

# Generate image
prompt = '''Now I want you to show a spaceships approaching Andromeda Galaxy!, it should be futuristic spaceships out on exploring universe'''
negative_prompt = " " # using an empty string if you do not have specific concept to remove

# Set image dimensions (16:9 aspect ratio)
width, height = 1664, 928

image = pipe(
    prompt=prompt + positive_magic["en"],
    negative_prompt=negative_prompt,
    width=width,
    height=height,
    num_inference_steps=50,
    true_cfg_scale=4.0,
    generator=torch.Generator(device="cuda").manual_seed(42)
).images[0]

# Save image to img folder
image.save("img/example_3.png")

# Display the image
display(image)
```

### Output

![Futuristic Spaceships Approaching Andromeda](/uploads/example_3.webp)

This demonstrates Qwen-Image's ability to create detailed sci-fi scenes.

## Advanced Features and Capabilities

### Flexible Aspect Ratios

Qwen-Image supports various aspect ratios for different use cases:

```python
aspect_ratios = {
    "1:1": (1328, 1328),    # Social media posts
    "16:9": (1664, 928),    # Widescreen displays
    "9:16": (928, 1664),    # Mobile/portrait content
    "4:3": (1472, 1140),    # Traditional displays
    "3:4": (1140, 1472),    # Portrait photos
    "3:2": (1584, 1056),    # Standard photos
    "2:3": (1056, 1584),    # Portrait standard
}
```

### Generation Parameters

Fine-tune your outputs with these key parameters:

-   `num_inference_steps`: Higher values (50-100) for better quality
-   `true_cfg_scale`: Controls prompt adherence (3.0-7.0 range)
-   `generator`: Set seed for reproducible results
-   `negative_prompt`: Specify elements to avoid

## Practical Applications

Qwen-Image's unique text rendering capabilities enable numerous applications:

-   **Marketing Materials**: Create posters, banners, and advertisements with perfectly rendered text
-   **Educational Content**: Generate infographics and diagrams with clear labels and annotations
-   **Social Media**: Design posts with integrated text overlays and captions
-   **Multilingual Content**: Produce visuals with text in multiple languages
-   **Creative Projects**: Combine artistic imagery with typographic elements
-   **UI/UX Mockups**: Generate interface designs with readable text elements

## Performance Optimization

To maximize performance on Vast.ai:

1. **Use bfloat16 precision**: Reduces memory usage while maintaining quality
2. **Batch generation**: Process multiple prompts together when possible
3. **Cache the model**: Keep it loaded between generations to avoid reload overhead
4. **Optimize resolution**: Start with lower resolutions for testing, then increase for final outputs

## Conclusion

Qwen-Image represents a significant advancement in image generation technology, particularly in its ability to seamlessly integrate readable text into visual content. Its support for multiple languages, mathematical notation, and various text styles makes it uniquely suited for projects requiring precise text-image integration.

By deploying on Vast.ai, you gain access to powerful GPUs capable of running this advanced model efficiently. Whether you're creating marketing materials, educational content, or artistic compositions, Qwen-Image provides the tools to generate professional-quality visuals with integrated text.

### Next Steps

Try these to explore further:

-   **Experiment with multilingual prompts**: Test the model's capabilities with Chinese, Japanese, or mixed-language text
-   **Create branded content**: Generate marketing materials with your company's text and logos
-   **Build an API service**: Wrap the model in a REST API for integration with your applications
-   **Explore style transfer**: Use the model's editing capabilities to transform existing images

[Get started on Vast.ai](https://cloud.vast.ai/) and experience Qwen-Image today!


Deploy Qwen-Image on Vast.ai to create stunning visuals with perfectly rendered text, multilingual support, and advanced editing features.

Deploying Qwen-Image for Advanced Text-Integrated Image Generation on Vast.ai


## Introduction

OpenAI has just released the GPT-OSS family of models, marking their return to open-weight model releases. The family includes GPT-OSS-120B (120 billion parameters) and GPT-OSS-20B (20 billion parameters), offering developers access to state-of-the-art language capabilities previously locked behind proprietary APIs.

GPT-OSS offers configurable reasoning capabilities with reasoning levels - from "low" for quick responses to "high" for complex problem-solving. The model uses OpenAI's harmony encoding system, which structures conversations and reasoning in a sophisticated format that enables fine-grained control over the model's thinking process.

This breakthrough means you can now:

-   Deploy OpenAI-quality models on your own infrastructure
-   Scale inference according to your exact needs

Vast.ai provides the perfect platform for running GPT-OSS models. With its marketplace of GPUs, you can choose the right hardware for your needs - whether running the efficient 20B model or the powerful 120B variant.

In this guide, we'll show you how to deploy both GPT-OSS models on Vast.ai using vLLM for optimized inference, with a focus on the 120B model. You'll learn how to interact with these models using the harmony encoding system for different reasoning levels.

## Setting Up the Environment

Before we can deploy our model, we need to set up our Vast.ai environment. First, install the Vast SDK:

```bash
# Install required packages
pip install --upgrade vastai
pip install --upgrade openai
pip install --upgrade openai-harmony
```

Set up your Vast API key (available from your [Account Page](https://cloud.vast.ai/account/)):

```bash
# Set your Vast.ai API key
export VAST_API_KEY="" # Your key here
vastai set api-key $VAST_API_KEY
```

## Choosing the Right Hardware

The GPT-OSS models have different hardware requirements:

-   **GPT-OSS-120B**: Requires an H100 GPU with 80GB VRAM
-   **GPT-OSS-20B**: Requires 16GB+ VRAM

For this guide, we'll demonstrate with the 120B model on H100. Let's search for suitable instances:

```bash
# Search for suitable GPU instances
vastai search offers " \
gpu_name = H100_SXM \
geolocation=US \
static_ip = true \
direct_port_count >= 1 \
verified = true \
disk_space >= 120 \
rentable = true"
```

## Deploying GPT-OSS with vLLM

Now let's deploy our GPT-OSS instance using vLLM's optimized inference server. We'll use the latest vLLM image with special GPT-OSS support. The pip install commands are from the model's Hugging Face pages ([120B](https://huggingface.co/openai/gpt-oss-120b), [20B](https://huggingface.co/openai/gpt-oss-20b)).

### Deploying GPT-OSS-120B (H100 recommended)

```bash
# Deploy vLLM instance for 120B model
export INSTANCE_ID= # Insert instance ID from search results

vastai create instance $INSTANCE_ID --image vllm/vllm-openai:latest --env '-p 8000:8000' --disk 160  --onstart-cmd 'uv pip install --system --upgrade transformers kernels torch openai; uv pip install --system --pre vllm==0.10.1+gptoss --extra-index-url https://wheels.vllm.ai/gpt-oss/ --extra-index-url https://download.pytorch.org/whl/nightly/cu128; vllm serve openai/gpt-oss-120b --max-model-len 80000'
```

### Alternative: Deploying GPT-OSS-20B on A100

Here's how to deploy the 20B model on an A100 GPU:

```bash
# Deploy vLLM instance for 20B model
export INSTANCE_ID= # Insert instance ID from search results

vastai create instance $INSTANCE_ID --image vllm/vllm-openai:latest --env '-p 8000:8000' --disk 80  --onstart-cmd 'uv pip install --system --upgrade transformers kernels torch openai; uv pip install --system --pre vllm==0.10.1+gptoss --extra-index-url https://wheels.vllm.ai/gpt-oss/ --extra-index-url https://download.pytorch.org/whl/nightly/cu128; vllm serve openai/gpt-oss-20b'
```

After deployment, wait for the model to download and start serving. You can monitor the logs to see when it's ready. Once running, find your instance's IP address and port from the [Instances tab](https://cloud.vast.ai/instances/) in the Vast AI Console.

## Interacting with GPT-OSS Using Harmony Encoding

GPT-OSS uses OpenAI's harmony encoding system, which provides structured conversation formatting and enables different reasoning levels. Let's set up our client:

```python
from openai import OpenAI
from openai_harmony import (
    load_harmony_encoding,
    HarmonyEncodingName,
    Role,
    Message,
    Conversation,
    SystemContent,
    DeveloperContent,
)

# Your server details
VAST_IP_ADDRESS = ""
VAST_PORT = ""

# Initialize client
client = OpenAI(
    api_key="EMPTY",
    base_url=f"http://{VAST_IP_ADDRESS}:{VAST_PORT}/v1"
)

# Load harmony encoding
enc = load_harmony_encoding(HarmonyEncodingName.HARMONY_GPT_OSS)

## Simple Chat Function
def chat_gpt_oss(prompt, reasoning="medium"):
    """
    Simple chat with GPT-OSS
    reasoning: "low", "medium", or "high"
    """
    # Create conversation with harmony format
    conversation = Conversation.from_messages([
        Message.from_role_and_content(Role.SYSTEM, SystemContent.new()),
        Message.from_role_and_content(
            Role.DEVELOPER,
            DeveloperContent.new().with_instructions(f"Reasoning: {reasoning}")
        ),
        Message.from_role_and_content(Role.USER, prompt),
    ])

    # Get formatted prompt
    formatted_prompt = enc.render_conversation_for_completion(conversation, Role.ASSISTANT)

    # Make API call
    response = client.completions.create(
        model="openai/gpt-oss-120b",
        prompt=formatted_prompt,
        max_tokens=2000,
        temperature=0.7,
    )

    return response.choices[0].text


prompt = "Explain quantum computing to a 4th grader. Keep in concise."
```

## Comparing Reasoning Levels

One of GPT-OSS's most interesting features is its configurable reasoning levels. Let's see how the model responds differently to the same prompt with low and high reasoning:

### Low Reasoning Response

```python
result = chat_gpt_oss(prompt, "low")
print("----LOW REASONING----")
print(result)
```

Output:

```
----LOW REASONING----
analysisWe need to explain quantum computing to a 4th grader, concise. Use simple analogies, like bits as light switches, qubits as coins that can be heads, tails, or both. Keep short.assistantfinal**Quantum Computing in a Nutshell (for a 4th‑grader)**

1. **Regular computers = light switches**
   - A normal computer works with bits that are either **0** (off) or **1** (on), just like a light switch that's either down or up.

2. **Quantum computers = magic coins**
   - A quantum computer uses **qubits**.
   - Think of a qubit like a coin that can be **heads (0)**, **tails (1)**, **or both at the same time** while it's spinning in the air. That "both at once" is called **superposition**.

3. **Working together**
   - When you have many regular bits, you have to try every possible combination one after another.
   - With many qubits, they can be in many combinations **all at once** because they can be linked together in a special way called **entanglement**. It's like a group of magic coins that always know what the others are doing, even if they're far apart.

4. **Why it matters**
   - Because of superposition and entanglement, a quantum computer can solve some puzzles (like cracking codes or simulating molecules) **much faster** than a regular computer.

**Bottom line:** A regular computer is like a row of on/off switches; a quantum computer is like a bunch of spinning magic coins that can be many things at once, letting it try many solutions at the same time.
```

### High Reasoning Response

```python
result = chat_gpt_oss(prompt, "high")
print("----HIGH REASONING----")
print(result)
```

Output:

```
----HIGH REASONING----
analysisWe need to explain quantum computing to a 4th grader, concise. Use simple analogies, avoid jargon, but keep accurate. Should be short. Provide analogies like coins, dice, superposition, entanglement. Keep concise.assistantfinal**What is a quantum computer?**

Imagine you have a super‑smart robot that can solve puzzles much faster than a regular computer. A quantum computer is that robot, but instead of using normal "on/off" switches (bits) like a regular computer, it uses tiny, magical switches called **qubits**.

### Bits vs. Qubits
- **Bit (regular computer):** Like a tiny light that can be **off (0)** or **on (1)**.
- **Qubit (quantum computer):** Like a tiny spinning top that can be **0**, **1**, **or both at the same time** (this is called *superposition*).

### Why is that cool?
If you have 10 bits, they can show only one of 2¹⁰ = 1,024 possible patterns at a time.
If you have 10 qubits, they can be in all 2¹⁰ = 1,024 patterns **at once**! That lets a quantum computer try many solutions together.

### A Simple Analogy
Think of a maze puzzle:

- **Regular computer:** Sends a single mouse down the maze, step by step, until it finds the exit.
- **Quantum computer:** Sends a whole swarm of mice that can explore every possible path at the same time and tell you the exit almost instantly.

### Entanglement (Teamwork)
Sometimes two qubits become "best friends" and link together, no matter how far apart they are. Changing one instantly changes the other. This is called **entanglement**, and it lets quantum computers coordinate their work in a super‑efficient way.

### Bottom Line
A quantum computer uses qubits that can be 0, 1, or both at once, and they can be entangled with each other. This lets the machine look at many possibilities at the same time, solving certain problems way faster than ordinary computers.

So, it's like a super‑fast, super‑clever puzzle‑solver that can try lots of answers all at once!
```

## Comparing Low vs High Reasoning Outputs

The quantum computing explanations demonstrate the distinct capabilities of GPT-OSS-120B's reasoning levels. With low reasoning, the model produced a concise, structured explanation using simple analogies like "magic coins" and "light switches" - delivering clear concepts in a compact format suitable for quick understanding.

The high reasoning response revealed significantly more sophisticated thinking. The model provided detailed analogies like the maze puzzle with mice, explained mathematical concepts (2¹⁰ patterns), and structured the explanation with clear sections and formatting. Most notably, it offered multiple perspectives on the same concepts - from the spinning top analogy to the teamwork explanation of entanglement.

This difference in reasoning depth represents a significant advancement in model control. Rather than simply adjusting temperature or other parameters, GPT-OSS allows direct specification of how thoroughly the model should think through problems. The low reasoning approach prioritized efficiency while maintaining accuracy, while high reasoning prioritized comprehensive understanding with rich examples and detailed explanations.

## Conclusion

OpenAI's first open-source model thus introduces not just accessibility to advanced language capabilities, but a new paradigm for controlling model reasoning depth - enabling applications to dynamically adjust between quick responses and thorough analysis based on context and requirements.


Learn how to deploy and run OpenAI's GPT-OSS models on Vast.ai using vLLM, with configurable reasoning levels and harmony encoding system.

Running OpenAI's GPT-OSS on Vast.ai


The DeepSeek R1-0528 release introduces a significant improvement to the reasoning capabilities of language models by making it easier to access "thinking" mode without requiring complex prompt engineering or pre-pending thinking tokens. This enhanced model provides transparent, step-by-step reasoning that's particularly valuable for educational applications, complex problem solving, and scenarios where transparency in AI decision-making is crucial.

In this post, we'll explore how to deploy the **DeepSeek-R1-0528-Qwen3-8B** model using **vLLM** on Vast.ai's cloud GPU platform, leveraging the new **qwen3 reasoning parser** that simplifies access to the model's internal thinking process.

---

## The Open Source AI Revolution

DeepSeek R1-0528 represents more than just a technical advancement—it's a significant step toward democratizing advanced AI capabilities. As an **open source model**, it makes these reasoning and advanced AI capabilities more accessible.

Unlike closed-source models that require vendor dependence, usage costs, and data privacy concerns, DeepSeek R1-0528 enables **complete control** over your AI infrastructure. Organizations can now deploy advanced reasoning capabilities locally, maintaining data sovereignty while avoiding variable per-token pricing.

This democratization means organizations can access cutting-edge AI without massive infrastructure investments. Advanced reasoning capabilities are no longer exclusive to large tech companies—anyone with GPU access can deploy their own reasoning-capable AI system.

---

## What Makes DeepSeek R1-0528 Special?

The DeepSeek R1-0528 release addresses a key limitation in previous DeepSeek models: the complexity of accessing the model's internal reasoning process. Previously, users needed to manually prepend "thinking" tokens to model outputs or use complex prompt engineering to see step-by-step reasoning.

---

## Deploying DeepSeek R1-0528 on Vast.ai

### Step 1: Set Up Vast.ai Environment

First, install the Vast CLI and configure your API key:

```bash
pip install --upgrade vastai

export VAST_API_KEY="your_vast_api_key"
vastai set api-key $VAST_API_KEY
```

### Step 2: Find Suitable Hardware

The DeepSeek-R1-0528-Qwen3-8B model requires specific hardware capabilities:

-   **Minimum 24GB GPU RAM** for model weights and KV cache
-   **Single GPU configuration** (sufficient for the 8B parameter model)
-   **Static IP address** for stable API endpoint hosting
-   **Direct port access** for the vLLM OpenAI-compatible server
-   **60GB+ disk space** for model storage and dependencies

Search for suitable instances:

```bash
vastai search offers "compute_cap >= 750 \
gpu_ram >= 24 \
num_gpus = 1 \
static_ip = true \
direct_port_count >= 1 \
verified = true \
disk_space >= 60 \
rentable = true"
```

### Step 3: Deploy the Model

Deploy using the vLLM OpenAI-compatible server with the qwen3 reasoning parser:

```bash
export INSTANCE_ID="your_instance_id"

vastai create instance $INSTANCE_ID \
--image vllm/vllm-openai:latest \
--env '-p 8000:8000' \
--disk 60 \
--args --model deepseek-ai/DeepSeek-R1-0528-Qwen3-8B \
--max-model-len 4096 \
--reasoning-parser qwen3
```

**Key deployment parameters:**

-   `--reasoning-parser qwen3`: Enables the simplified reasoning parser
-   `--max-model-len 4096`: Accommodates longer reasoning sequences
-   `--tensor-parallel-size 1`: Single GPU configuration
-   `--gpu-memory-utilization 0.90`: Optimized memory usage

### Step 4: Get Instance Details and Test Connection

Navigate to the [Instances tab](https://cloud.vast.ai/instances/) in the Vast AI Console and locate your instance. Click the IP address button to view the forwarded ports.

Test the connection with a simple curl command:

```bash
export VAST_IP_ADDRESS="your_ip_address"
export VAST_PORT="your_port"

curl -X POST http://$VAST_IP_ADDRESS:$VAST_PORT/v1/completions \
-H "Content-Type: application/json" \
-d '{"model": "deepseek-ai/DeepSeek-R1-0528-Qwen3-8B", "prompt": "Hello, how are you?", "max_tokens": 50}'
```

---

## Testing the Model's Reasoning Capabilities

### Setup Python Client

```python
import time
from openai import OpenAI

# Replace with your actual Vast.ai instance details
VAST_IP_ADDRESS = "your_ip_address"
VAST_PORT = "your_port"

client = OpenAI(
    api_key="DUMMY",
    base_url=f"http://{VAST_IP_ADDRESS}:{VAST_PORT}/v1"
)
```

### Basic Functionality Test

```python
start = time.time()
test_resp = client.chat.completions.create(
    model="deepseek-ai/DeepSeek-R1-0528-Qwen3-8B",
    messages=[{"role": "user", "content": "Hello"}]
)
elapsed = time.time() - start

print("Response:", test_resp.choices[0].message.content)
print(f"Elapsed time: {elapsed:.4f} s")
```

### Reasoning Capability Test

This is where DeepSeek R1-0528 truly shines. The model naturally provides step-by-step reasoning without complex prompt engineering:

```python
reasoning_prompt = """
Write the History of Paris in a paragraph, show your thinking
step by step first before paragraph and then write the history
in a paragraph
"""

start = time.time()
resp = client.chat.completions.create(
    model="deepseek-ai/DeepSeek-R1-0528-Qwen3-8B",
    messages=[{"role": "user", "content": reasoning_prompt}]
)
elapsed = time.time() - start

print("Model output:\n", resp.choices[0].message.content)
print(f"Elapsed time: {elapsed:.3f} s")
```

**Expected output structure:**

1. **Thinking process**: The model shows its step-by-step reasoning
2. **Final answer**: A well-structured paragraph about Paris's history

### Mathematical Reasoning Test

```python
math_prompt = "If 8 × 7 = ?, show your work step by step."

resp = client.chat.completions.create(
    model="deepseek-ai/DeepSeek-R1-0528-Qwen3-8B",
    messages=[{"role": "user", "content": math_prompt}]
)

print("Mathematical reasoning:\n", resp.choices[0].message.content)
```

The model will naturally break down the problem, show intermediate steps, and arrive at the solution with clear reasoning traces.

---

## Key Benefits of DeepSeek R1-0528

### 1. **Simplified Reasoning Access**

No need to prepend thinking tokens or use complex prompt engineering. The `qwen3` reasoning parser automatically handles the extraction and presentation of reasoning steps.

### 2. **Educational Value**

Perfect for educational applications where students need to understand the problem-solving process, not just the final answer.

### 3. **Transparency and Trust**

Users can see exactly how the model arrives at its conclusions, building trust in AI-powered decision-making systems.

### 4. **Cost-Effective Deployment**

Running on Vast.ai provides significant cost savings compared to major cloud providers while maintaining high performance.

### 5. **Easy Integration**

OpenAI-compatible API ensures seamless integration with existing applications and workflows.

---

## Use Cases

### Educational Applications

-   Step-by-step problem solving for mathematics, science, and logic
-   Teaching critical thinking and reasoning skills
-   Creating interactive learning experiences

### Business Intelligence

-   Transparent decision-making in automated systems
-   Audit trails for AI-powered recommendations
-   Complex analysis with clear reasoning paths

### Research and Development

-   Understanding model behavior and capabilities
-   Testing reasoning patterns across different domains
-   Developing more interpretable AI systems

---

## Conclusion

The DeepSeek R1-0528 release represents a significant step forward in making AI reasoning more accessible and transparent. By combining this powerful model with Vast.ai's cost-effective GPU infrastructure and the simplified qwen3 reasoning parser, developers can now easily deploy transparent, reasoning-capable AI systems.

Whether you're building educational tools, business intelligence systems, or research applications, DeepSeek R1-0528 on Vast.ai provides an excellent foundation for applications that require clear, step-by-step reasoning capabilities.

The elimination of complex prompt engineering for accessing reasoning modes makes this model particularly valuable for production applications where transparency and interpretability are key requirements.

---

## Useful Resources

-   [DeepSeek R1-0528 Model on Hugging Face](https://huggingface.co/deepseek-ai/DeepSeek-R1-0528-Qwen3-8B)
-   [vLLM Documentation](https://docs.vllm.ai/)
-   [Vast.ai GPU Marketplace](https://vast.ai)
-   [OpenAI Python SDK](https://github.com/openai/openai-python)

---

_Ready to deploy transparent, reasoning-capable AI? Get started with DeepSeek R1-0528 on Vast.ai today! 🚀_


Deploy DeepSeek R1-0528 on Vast.ai to access transparent, step-by-step AI reasoning without prompt tricks—ideal for education, research, and business use.

DeepSeek R1-0528: Enhanced Reasoning with Simplified Thinking Mode on Vast.ai


Ten years ago today, our CEO and founder, Jacob Cannell, published ["The Brain as a Universal Learning Machine"](https://www.lesswrong.com/posts/9Yc7Pp7szcjPgPsjf/the-brain-as-a-universal-learning-machine) on LessWrong.com.

Drawing from neuroscience, machine learning, and systems theory, he proposed that **the human brain is not a patchwork of specialized, genetically programmed modules, but a general-purpose learning system** – or, as he put it, "a biological implementation of a _Universal Learning Machine_."

In hindsight, the timing was notable. Cannell published the piece just as deep learning was moving from niche research to broader relevance in artificial intelligence (AI). In the decade since then, rapid progress in scalable, trainable, data-driven models has validated the idea that learning, not pre-engineered structure, is the key to building intelligent systems.

So on the tenth anniversary of its publication, we'd like to revisit the key ideas of the article, reflect on what's aged well, and look ahead at what's still unfolding.

We’d encourage you to read the full article on [Less Wrong](https://www.lesswrong.com/posts/9Yc7Pp7szcjPgPsjf/the-brain-as-a-universal-learning-machine).


A look at the brain as a universal learning machine, and how it can be used to learn about the world.

A Decade Later: Revisiting the Brain as a Universal Learning Machine


Google has recently released Gemma 3, their latest open-source language model family. The model, designed to run efficiently on a single GPU, brings state-of-the-art capabilities including advanced multimodal understanding, multilingual support, and an impressive 128K token context window. In this post, we'll explore the key features of Gemma 3 and demonstrate how to deploy it on Vast.ai to leverage one of its new features, comparing images.

## What makes Gemma 3 Special

Gemma 3 is available in four model sizes: 1B, 4B, 12B, and 27B parameters. Each size comes in both instruction-tuned (IT) and pre-trained (PT) variants, allowing developers to choose the optimal model for their specific applications.

What makes Gemma 3 particularly accessible is its efficient architecture - all models are optimized to run on a single GPU, with the smaller variants capable of running on consumer-grade GPUs through Vast.ai's platform.

Gemma 3 offers multiple state-of-the-art capabilities:

### 1. Enhanced Context Window

-   Supports up to 128K tokens context window
-   Improved long-range understanding and coherence
-   Better handling of complex, multi-turn conversations

### 2. Multi-Modal Capabilities

-   Native image understanding and analysis (available in 4B, 12B, and 27B models only)
-   Support for both text and image inputs in the same prompt
-   Advanced ability to analyze images, text, and short videos

### 3. Improved Performance

-   State-of-the-art performance for its size
-   Enhanced reasoning capabilities
-   Better code generation and understanding

### 4. Multilingual Support

-   Out-of-the-box support for over 35 languages
-   Pretrained support for over 140 languages

### 5. Additional Features

-   Function calling support for automated workflows
-   Official quantized versions for faster performance
-   Optimized for various hardware platforms (NVIDIA GPUs, Google Cloud TPUs, AMD GPUs)

## Using Gemma 3 For Image Comparison On Vast

Gemma 3 supports multimodal inputs, including an impressive image comparison capability. In this guide, we'll demonstrate this feature using images from the Open Image Preferences dataset, available [here](https://huggingface.co/datasets/data-is-better-together/open-image-preferences-v1-binarized).

### Image Selection

For this post, we've selected two contrasting astronaut images from the Open Image Preferences dataset:

<div style={{ display: "flex", gap: "20px", marginBottom: "2rem" }}>
    <div style={{ flex: 1 }}>
        <img
            src="/uploads/image-1.jpg"
            alt="Astronaut 1"
            style={{ width: "100%" }}
        />
        <p style={{ textAlign: "center", fontWeight: "bold" }}>Astronaut 1</p>
    </div>
    <div style={{ flex: 1 }}>
        <img
            src="/uploads/image-2.jpg"
            alt="Astronaut 2"
            style={{ width: "100%" }}
        />
        <p style={{ textAlign: "center", fontWeight: "bold" }}>Astronaut 2</p>
    </div>
</div>

### Launch an Instance

First, let's set up a Vast.ai instance with the necessary configurations to run Gemma 3:

```bash
vastai create instance $INSTANCE_ID \
    --image vllm/vllm-openai:latest \
    --disk 40 \
    --env '-p 8000:8000 -e HUGGING_FACE_HUB_TOKEN=hf-token' \
    --args --model google/gemma-3-4b-it --gpu-memory-utilization 0.9 --max-model-len 4096 --limit-mm-per-prompt "image=2"
```

### Call Gemma 3 Using the OpenAI SDK

Next, we'll write a script to call Gemma 3 on our instance. vLLM exposes an OpenAI-compatible API endpoint, allowing us to interact with Gemma 3 using the OpenAI SDK:

```python
from openai import OpenAI
import base64

# Your Vast.ai instance details
VAST_IP_ADDRESS = ""
VAST_PORT = ""

# Initialize the client
client = OpenAI(
    api_key="EMPTY",
    base_url=f"http://{VAST_IP_ADDRESS}:{VAST_PORT}/v1"
)

# Load and convert local images to base64
def image_to_base64(image_path):
    with open(image_path, "rb") as image_file:
        return base64.b64encode(image_file.read()).decode('utf-8')

# Convert images to base64
image1_base64 = image_to_base64("/path/to/image-1.jpg")
image2_base64 = image_to_base64("/path/to/image-2.jpg")

response = client.chat.completions.create(
    model="google/gemma-3-4b-it",
    messages=[
        {
            "role": "user",
            "content": [
                {
                    "type": "text",
                    "text": "Compare these two images"
                },
                {
                    "type": "image_url",
                    "image_url": {
                        "url": f"data:image/jpeg;base64,{image1_base64}"
                    }
                },
                {
                    "type": "image_url",
                    "image_url": {
                        "url": f"data:image/jpeg;base64,{image2_base64}"
                    }
                }
            ]
        }
    ]
)

print(response.choices[0].message.content)
```

Here is our output from Gemma 3:

```
Okay, let's compare these two images of astronauts. Here's a breakdown of the key differences and similarities:

**Similarities:**

* **Subject:** Both images feature a spacesuit and the astronaut's helmet.
* **Style:**  Both are highly stylized and appear to be digitally rendered images, likely created with a focus on aesthetic appeal and a sense of wonder. They evoke a futuristic and somewhat surreal feeling.
* **Helmet:** The overall shape and design of the helmet are fairly consistent between the two – a classic, bulbous, and slightly vintage-looking design.



**Differences:**

1. **Background:** This is the most significant difference.
   * **Image 1:** Features a vibrant, complex, and detailed space scene with a blue and gold starry sky, nebulas, and a visible horizon line suggesting a planet or moon. The background is rich in color and texture.
   * **Image 2:** Shows a much simpler and muted background. It’s a dusky blue sky with scattered stars and a faint suggestion of a horizon with mountains. There's no nebula or dramatic planetary features.

2. **Suit Detail & Appearance:**
    * **Image 1:** The spacesuit is more elaborate and brightly colored. It has intricate gold detailing on the arms and chest and a much more modern and polished look. It also appears to be a little more reflective.
    * **Image 2:** The suit is weathered and looks older. There's a visible rust and decay on the metal parts, suggesting it's been used and exposed to the elements. It’s a more rugged and worn appearance.

3. **Pose and Composition:**
   * **Image 1:** The astronaut is standing and facing slightly towards the camera, but the composition is broader, showing more of the suit and the expansive space.
   * **Image 2:** The astronaut is positioned at a three-quarter angle, focusing primarily on the helmet and upper body. The composition is more intimate and tightly framed.

4. **Lighting & Tone:**
   * **Image 1:**  The lighting is dramatic and highlights the spacesuit with warm, glowing lights. It's a brighter and more vibrant image overall.
   * **Image 2:** The lighting is softer and cooler, creating a more subdued and melancholic atmosphere.

**In essence:**

* **Image 1** is a grand, visually stunning depiction of an astronaut experiencing the majesty of space.
* **Image 2** is a more intimate and evocative portrayal, suggesting a sense of isolation, exploration, and the passage of time.



Would you like me to elaborate on any specific aspect of the comparison, such as the artistic style or the possible symbolism of the images?
```

### Image Comparison Analysis

The model provides detailed, structured analysis of the images, demonstrating its sophisticated visual understanding capabilities. The analysis includes:

1. **Comprehensive Comparison**: The model systematically breaks down both similarities and differences between the images
2. **Structured Output**: Results are organized into clear categories (Similarities, Differences, In essence)
3. **Detailed Observations**: The model captures both obvious and subtle details, from overall composition to specific visual elements
4. **Contextual Understanding**: The analysis includes interpretation of artistic style, mood, and potential symbolism
5. **Interactive Engagement**: The model offers to elaborate on specific aspects of the comparison

This level of analysis showcases Gemma 3's advanced multimodal capabilities, particularly its ability to:

-   Process and compare multiple images simultaneously
-   Provide nuanced visual analysis
-   Generate structured, coherent responses
-   Maintain context throughout the analysis

## Next Steps

The image comparison capabilities we've demonstrated can be applied to several real-world use cases:

-   **Visual Quality Control**: Compare product images for manufacturing defects
-   **Design Evolution**: Analyze changes between design iterations
-   **Medical Imaging**: Compare diagnostic images for changes or anomalies

### Beyond Image Analysis: Additional Capabilities

While image comparison showcases Gemma 3's visual understanding, the model's versatility extends far beyond visual tasks. Here are several other powerful applications you can build:

-   **Multilingual Chatbots**: Create AI assistants that can communicate in 140+ languages
-   **Document Analysis**: Leverage the 128K token context window for processing extensive documents
-   **Educational Tools**: Build interactive learning systems with multilingual support
-   **Visual Search**: Combine text and image understanding for sophisticated search applications

These applications can be deployed using the same Vast.ai infrastructure we've explored, allowing for efficient development and scaling.

## Conclusion

Gemma 3 represents a significant advancement in open-source AI models, offering state-of-the-art performance while being designed to run efficiently on a single GPU. Vast.ai's infrastructure provides an ideal platform for deploying Gemma 3 in production environments, making it accessible for both development and scaling applications.


Google’s open-source Gemma 3 model brings multimodal reasoning, 128K context, and multilingual support—all runnable on a single GPU. Deploy it on Vast.ai to compare images or build advanced AI apps.

Posts about: AI

Report: NVIDIA Developing New Blackwell-Based AI Chip for China

Deploying Qwen-Image for Advanced Text-Integrated Image Generation on Vast.ai

Running OpenAI's GPT-OSS on Vast.ai

DeepSeek R1-0528: Enhanced Reasoning with Simplified Thinking Mode on Vast.ai

A Decade Later: Revisiting the Brain as a Universal Learning Machine

Running Google's Gemma 3 on Vast.ai