InferX Catalog | Kimi-VL-A3B-Thinking-2506

Kimi-VL-A3B-Thinking-2506

Kimi-VL-A3B-Thinking-2506 is a state-of-the-art vision-language model (VLM) released by Moonshot AI in mid-2025.

moonshotai multimodal image2text

Log in to deploy

Metadata

Provider

moonshotai

Modality

multimodal

API type

image2text

Source

huggingface / moonshotai/Kimi-VL-A3B-Thinking-2506

Created

2026-04-07 20:54:43 UTC

Updated

2026-04-13 03:08:48 UTC

Catalog version

2

Visibility

Published

Specifications

Parameters

16.00B

MoE

No

Max model length

2000

Image

vllm/vllm-openai:v0.16.0

Default Deploy Config

GPU count

1

vRAM

60000 MB

Summary

1xGPU 60000 MB

Recommended Use Cases

—

Model Spec

{
    "image": "vllm/vllm-openai:v0.16.0",
    "commands": [
        "--model",
        "moonshotai/Kimi-VL-A3B-Thinking-2506",
        "--trust-remote-code",
        "--max-model-len",
        "2000",
        "--tensor-parallel-size=1"
    ],
    "resources": {
        "GPU": {
            "Count": 1,
            "vRam": 60000
        }
    },
    "envs": [],
    "sample_query": {
        "body": {
            "max_tokens": "200",
            "temperature": "0"
        },
        "path": "v1/chat/completions",
        "prompt": "What is in this image?",
        "apiType": "image2text",
        "dataUrl": "https://www.ilankelman.org/stopsigns/australia.jpg",
        "prompts": [],
        "loadingTimeout": 90
    }
}