Ling 3.0 Flash VL
Neural Network
Ling 3.0 Flash VL is an InclusionAI multimodal MoE model: understands images and text, available via API on BotHub.
Specifications
- Input
- text, images, documents
- Output
- text
- Context size
- 131 072 tokens
- Max answer length
- 32 768 tokens
- Reasoning effort
- low, medium, high
- Capabilities
- reasoning, function calling
Prompt cost
(per 1M tokens)
Answer cost
(per 1M tokens)
How it works Ling 3.0 Flash VL?
Ling 3.0 Flash VL is an evolution of InclusionAI's Ling 3.0 Flash: the same 124B parameter MoE architecture with 5.5B active parameters, but with native visual perception and enhanced language capabilities. The model accepts images alongside text and analyzes what it sees in a general chat, without a separate service for images. Via BotHub, the model is accessible from Russia without a VPN or foreign card, and you pay for tokens in rubles—no subscription, and unused tokens do not expire. Over 250 models are available in one window: you can switch and compare answers, and a unified OpenAI-compatible API connects the model to your product, allowing you to replace it with another without rewriting the integration. Data is encrypted via AES-GCM and not stored, and companies have access to contracts, invoices, electronic document management, and an admin panel with limits. Useful when you need to describe image content, extract meaning from interface screenshots or diagrams, match an image with a text prompt, build a multimodal support assistant, or process a stream of visual materials via API.Frequently asked questions about Ling 3.0 Flash VL
You can use generated results for commercial purposes. You own all rights to the content you create. The only restriction: make sure your prompt does not include copyrighted third-party material. You are responsible for respecting the rights to any input data.
The model accepts text, documents, and images, and responds with text. It is suitable for analyzing files and screenshots, extracting data from tables and diagrams, working with code, analytics, and agent scenarios: it has a reasoning mode and function calling. Context is 131,072 tokens.
In a single response, the model outputs up to 32,768 tokens—enough for a detailed document analysis, a long article, or a large code snippet. If you need more text, break the task into parts: the 131,072-token input context allows you to keep the entire history nearby.
Yes, the model has a reasoning mode: before answering, it processes the task step-by-step. This significantly helps with math, logic, code debugging, and analyzing long documents. Note that reasoning consumes output tokens and slightly increases response time.
Yes, function calling and structured JSON output are supported. You can connect the model to your tools, databases, and external services, and receive responses strictly according to a schema. It works via the unified OpenAI-compatible BotHub API, so you can switch models later without rewriting the integration.
The model accepts images and documents—you can send a screenshot, diagram, scan, or PDF and ask it to analyze the content. There is no mention of audio or video in our data, so for such files, it is safer to use a specialized model: it is available in the same BotHub window.
There are no separate benchmarks for Russian in our data, so we won't make any promises here—evaluate the quality on your own tasks, it's quick. The BotHub interface is in Russian, access from Russia is available without a VPN or foreign card, payment is via Russian cards, and tokens do not expire.