inclusionAI: Ling 3.0 Flash VL

Ling 3.0 Flash VL extends the Ling 3.0 Flash lineage into a vision-language model that adds image and video understanding on top of the base text and reasoning capabilities. InclusionAI, with roots in Ant Group, positions the model as a multimodal variant of Ling 3.0 Flash, retaining the same sparse Mixture-of-Experts architecture with 124B total parameters and roughly 5.5B active per token, and adding a VideoRoPE component for temporal video understanding. The combination is meant to keep compute costs low while letting the model handle richer visual inputs than a text-only backbone could.

Practically, Ling 3.0 Flash VL targets visual comprehension, reasoning grounded in visual evidence, and interface interaction, with reported coverage of object counting, chart and document reading, and interface understanding for task automation. It supports both instant and reasoning modes and offers a large 262,144-token context window, making it well suited to long documents, multi-image workflows, and agent-style tasks that mix visual context with step-by-step thinking. Developers looking for a sparse-activation multimodal model from the Ling family will find it oriented toward grounded visual reasoning and automation rather than pure chat use.

Kilo Gatewayinclusionai/ling-3.0-flash-vlling

Open provider page Provider docs

Quick Info

Provider: Kilo Gateway
Model key: inclusionai/ling-3.0-flash-vl
Release date: Sep 10, 2026
Last updated: Sep 10, 2026
Input modalities
Output modalities
Capabilities

Cost

Input token cost: $0.06
Output token cost: $0.18

Limits

Output tokens: 32,768 tokens
Context window: 131,072 tokens

Transparent token rates

Compare inclusionAI: Ling 3.0 Flash VL pricing

Rates are shown per one million tokens. Combined means one million input plus one million output tokens.

Browse this family

Latest news about inclusionAI: Ling 3.0 Flash VL

Kilo Gateway

CoverageBenchmark

InclusionAI's Ling 3.0 Flash VL is a multimodal model that builds on Ling 3.0 Flash with a 124B-total / 5.5B-active Mixture-of-Experts architecture, adding native visual perception and advanced visual agent capabilities to the language foundation. It operates as a hybrid instant/reasoning model with tool calling, suppo Artificial Analysis benchmark scores published alongside the OpenRouter model page show Ling 3.0 Flash VL at an Intelligence Index of 24.6 (better than 45% of compared models), a Coding Index of 57.0, and an Agentic Index of 28.7. Per-evaluation results include 86.2% on GPQA Diamond for graduate-level scientific reason

Kilo Gateway

CoverageRelease Notes

An AI model release timeline compiled from the OpenRouter models API lists inclusionAI: Ling 3.0 Flash VL as an API launch dated 2026-09-10 on the Inclusionai API with a 131k context window. The entry appears among a broader daily-refreshed feed of model launches across providers including Xiaomi, SpaceXAI, PrismML, Z. The timeline entry adds no architectural, capability, or benchmark detail beyond what the model-specific OpenRouter page reports, functioning as an aggregator confirmation of the September 10, 2026 launch date and the 131k context size. No creator-side (InclusionAI) primary announcement or independent press coverage of

Kilo Gateway

CoverageRelease Notes

Artificial Analysis published language model evaluation results for InclusionAI's Ling-3.0-flash-VL on 10 September 2026, as recorded in its public changelog. The model is credited to InclusionAI and assigned an Intelligence Index score of 25, making it the only supplied candidate that explicitly names this exact varia The Intelligence Index 25 result is reported under the Artificial Analysis Intelligence Index methodology, which as of 7 September 2026 moved to v4.3 with Terminal-Bench upgraded to v4 and τ³-Banking replaced by AutomationBench-AA (657 business workflows on Zapier's private question set). This methodology context is re