InternVL Models by Intern-AI

Intern-AIApache 2.0Open source
Intern-AI releases · 1 in the last 12 monthsChangelog →
2 models2024Up to 32K ctx

Last refreshed 2026-05-22. Next refresh: weekly.

Details

ResearcherIntern-AI
LicenseApache 2.0OSI-approved
Commercial usePermitted
Models2
Released2024
Max context32K

About

InternVL is OpenGVLab’s open-source multimodal large language model family (InternVL-Chat), built to narrow the gap with commercial multimodal models. InternVL 1.5 uses a strong InternViT-6B vision encoder connected to an LLM through an MLP, dynamic high-resolution image tiling (up to 4K at test time), and a high-quality bilingual English–Chinese dataset aimed at OCR and Chinese-related tasks. The series continues through later InternVL 2.x/3.x releases on the same OpenGVLab project.

Decision facts

Best fit
coding
Capability starting point
InternVL 2 with 32K context
Lowest tracked input
Not tracked
Closest related family
Sage S2

Current Variants

Use-when guidance is based on each model's tracked capabilities, context window, release date, and replacement status.

2 in view
InternVL 2Current

Use when the workload needs 32K context.

2024-0832K context

Use when the workload needs 32K context and 26B parameters.

2024-0332K context26B parameters

Release Timeline

2 release groups
2024-08
1 current
InternVL 2
32K context
Current
2024-03
1 current
InternVL 1.5
32K context26B parameters
Current

Specifications(2 models)

InternVL model specifications comparison
ModelReleasedContextParameters
InternVL 22024-0832K—
InternVL 1.52024-0332K26B