Instructions to use Aidev2006/Inquisitive-V1-Pro-RL with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Aidev2006/Inquisitive-V1-Pro-RL with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="Aidev2006/Inquisitive-V1-Pro-RL", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("Aidev2006/Inquisitive-V1-Pro-RL", trust_remote_code=True, device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Aidev2006/Inquisitive-V1-Pro-RL with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Aidev2006/Inquisitive-V1-Pro-RL"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Aidev2006/Inquisitive-V1-Pro-RL",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Aidev2006/Inquisitive-V1-Pro-RL

SGLang

How to use Aidev2006/Inquisitive-V1-Pro-RL with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "Aidev2006/Inquisitive-V1-Pro-RL" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Aidev2006/Inquisitive-V1-Pro-RL",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "Aidev2006/Inquisitive-V1-Pro-RL" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Aidev2006/Inquisitive-V1-Pro-RL",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Aidev2006/Inquisitive-V1-Pro-RL with Docker Model Runner:
```
docker model run hf.co/Aidev2006/Inquisitive-V1-Pro-RL
```

SYNIN V1.0 PRO

A multimodal foundation model for reasoning, coding, agents, and long-context intelligence.

1.02T Total Parameters • 42B Activated Parameters • 1M Context • 4 Modalities

SYNIN AI • Website

SYNIN V1.0 PRO

SYNIN V1.0 PRO is the flagship model release in the SYNIN AI model family.

It is designed for advanced reasoning, coding, agentic workflows, multimodal understanding, and long-context applications.

SYNIN V1.0 PRO is based on Xiaomi MiMo-V2.6-Pro-RL and is maintained as an independently versioned SYNIN model repository.

Provenance

This model is based on XiaomiMiMo/MiMo-V2.6-Pro-RL. The underlying model weights and architecture originate from the upstream MiMo release.

Model Overview

Property	Specification
Model	SYNIN V1.0 PRO
Base Model	`XiaomiMiMo/MiMo-V2.6-Pro-RL`
Architecture	Sparse Mixture-of-Experts
Total Parameters	1.02T
Activated Parameters	42B
Context Length	1M tokens
Modalities	Text, Image, Video, Audio
Layers	70
Routed Experts	384
Expert Routing	Top-8
License	MIT
Library	Transformers

Capabilities

Reasoning

SYNIN V1.0 PRO is designed for complex, multi-step reasoning and problem solving.

Mathematical reasoning
Logical reasoning
Analysis
Planning
Multi-step problem solving
Long-context reasoning

Coding

Designed for software engineering and technical workflows.

Code generation
Code explanation
Debugging
Refactoring
Technical analysis
Agentic coding workflows

Multimodal Understanding

The model supports multimodal inputs through its underlying architecture.

Text
Images
Video
Audio

Agents

Designed to serve as an intelligence layer for tool-using and agentic systems.

Tool use
Planning
Multi-step execution
Environment interaction
Agent workflows

Architecture

SYNIN V1.0 PRO uses a large-scale sparse Mixture-of-Experts architecture.

                         SYNIN V1.0 PRO
                                │
                                ▼
                       ┌─────────────────┐
                       │   Input Layer   │
                       └────────┬────────┘
                                │
              ┌─────────────────┼─────────────────┐
              │                 │                 │
              ▼                 ▼                 ▼
            Text              Vision            Audio
              │                 │                 │
              └─────────────────┼─────────────────┘
                                │
                                ▼
                       ┌─────────────────┐
                       │ Hybrid Attention│
                       └────────┬────────┘
                                │
                                ▼
                     ┌──────────────────────┐
                     │      Sparse MoE      │
                     │                      │
                     │  384 Routed Experts  │
                     │      Top-8 Routing   │
                     └──────────┬───────────┘
                                │
                                ▼
                       ┌─────────────────┐
                       │  Output Layer   │
                       └─────────────────┘

Core Architecture

70 transformer layers

1 dense layer

69 MoE layers

384 routed experts

Top-8 expert routing

Hybrid attention

Long-context architecture

Multimodal processing



---

1M Context

SYNIN V1.0 PRO supports a context window of up to:

<div align="center">1,000,000 TOKENS

</div>The extended context enables applications involving:

Large documents

Long conversations

Large codebases

Research material

Extended agent workflows

Multi-document analysis



---

SYNIN Intelligence Stack

SYNIN V1.0 PRO is designed to operate as the core intelligence layer of the broader SYNIN AI platform.

SYNIN AI
                            │
            ┌───────────────┼───────────────┐
            │               │               │
            ▼               ▼               ▼
       SYNIN CHAT      SYNIN VOICE     SYNIN VISION
            │               │               │
            └───────────────┼───────────────┘
                            │
                            ▼
                    SYNIN V1.0 PRO
                            │
             ┌──────────────┼──────────────┐
             │              │              │
             ▼              ▼              ▼
         Reasoning        Agents         Coding
             │              │              │
             └──────────────┼──────────────┘
                            │
                            ▼
                       SYNIN API


---

Intended Use

SYNIN V1.0 PRO is intended for research, development, and production applications including:

AI assistants

Conversational AI

Coding assistants

Autonomous agents

Research assistants

Document intelligence

Multimodal applications

Long-context applications

Developer platforms

AI APIs



---

Usage

Example loading pattern with Transformers:

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "YOUR_ORG/SYNIN-V1.0-PRO"

tokenizer = AutoTokenizer.from_pretrained(
    model_id,
    trust_remote_code=True
)

model = AutoModelForCausalLM.from_pretrained(
    model_id,
    trust_remote_code=True,
    device_map="auto"
)

Replace YOUR_ORG/SYNIN-V1.0-PRO with the actual Hugging Face repository containing the model.

For large-scale inference, use an inference engine and multi-GPU deployment configuration appropriate for the model architecture.


---

Deployment Architecture

Client
                           │
                           ▼
                       SYNIN API
                           │
                           ▼
                    Model Gateway
                           │
                 ┌─────────┴─────────┐
                 │                   │
                 ▼                   ▼
          SYNIN V1.0 PRO        Other Models
                 │
                 ▼
             GPU Cluster
                 │
          ┌──────┴──────┐
          │             │
          ▼             ▼
       Primary       Replica
       Cluster       Cluster

The SYNIN model gateway allows applications to route requests between different models and specialized inference systems.


---

Model Provenance

Base model:

XiaomiMiMo/MiMo-V2.6-Pro-RL

SYNIN V1.0 PRO is a SYNIN-branded derivative based on the upstream MiMo release.

This repository is maintained independently and does not automatically track future changes made to the upstream repository.

Future SYNIN releases may include additional post-training, optimization, alignment, evaluation, or proprietary model development.


---

Limitations

Like other large language and multimodal models, SYNIN V1.0 PRO may:

Generate incorrect information

Produce inaccurate reasoning

Misinterpret visual or audio inputs

Generate incorrect or insecure code

Reflect limitations in its training data

Produce outputs that require human verification


Applications should implement appropriate validation, monitoring, safety controls, and human oversight.


---

License

SYNIN V1.0 PRO is released under the MIT License, subject to the applicable terms and provenance of the underlying model.

See the LICENSE file in this repository for the complete license text.


---

Citation

@misc{synin-v1-pro,
  title        = {SYNIN V1.0 PRO},
  author       = {SYNIN AI},
  year         = {2026},
  publisher    = {SYNIN AI},
  note         = {Based on Xiaomi MiMo-V2.6-Pro-RL}
}


---

<div align="center">SYNIN AI

Intelligence, engineered.

<a href="https://synin.org">synin.org</a>

</div>
```0

Downloads last month: 217

Safetensors

Model size

1T params

Tensor type

BF16

F32

F8_E4M3

Model tree for Aidev2006/Inquisitive-V1-Pro-RL

Base model

XiaomiMiMo/MiMo-V2.6-Pro-RL

Quantized

(8)

this model