LLM Reference

MiniCPM-V 4.6

Released
2026-05-11
Last refreshed
2026-07-07
Status
Researched 6d ago
Open sourceCommercial use: permittedMultimodalRAGAgentsLong contextVisionJSON / Tool usemultimodalvision-language

MiniCPM-V 4.6 is a released rag, agents, and long context model with open-source and 262k context; evaluate it while provider pricing coverage matures.

Use it for

  • Teams evaluating rag, agents, and long context
  • Workloads that can use a 262k context window

Do not use it for

  • Cost-sensitive launches that need sourced token pricing
  • Teams that need a tracked hosted API route today
Specifications
Family
MiniCPM
Released
2026-05-11
Context
262k
Parameters
1.3B
Architecture
Decoder Only
Openness
Open source
License
Apache 2.0OSI-approvedCommercial use: permitted
Weights
Available
Code
Unknown
Created by

Efficient open-source language models for edge devices.

Beijing, China
Founded 2022
Website
Pricing

No tracked provider token pricing is available yet.

About

OpenBMB's compact 1.3B vision-language model, open-sourced May 11, 2026, for on-device deployment on smartphones (iOS, Android, HarmonyOS) and edge devices. It pairs a SigLIP2-400M vision encoder with a Qwen3.5-0.8B language backbone using the LLaVA-UHD v4 approach. It supports single-image, multi-image, and video input (up to 128 frames), tool-call-formatted text output, and a 262,144-token context window. OpenBMB reports a 13 score on the Artificial Analysis Intelligence Index and support for vLLM, SGLang, llama.cpp, Ollama, Transformers, SWIFT, and LLaMA-Factory. Apache 2.0 license.

MiniCPM-V 4.6 is an open-source model in the MiniCPM family. The structured metadata tracks a 262k-token context window, multimodal input, and tool use. Headline tracked benchmarks include MathVista 75.6 and Massive Multi-discipline Multimodal Understanding 55.3.

Top use-case fit: coding, agents, and build tasks

RAG

Included by capability and metadata signals in the decision map.

Agents

Included by capability and metadata signals in the decision map.

Long context

Included by capability and metadata signals in the decision map.

Provider price ladder

No tracked provider token pricing is available for this model yet.

Capabilities

VisionMultimodalTool Use

Benchmark peer barsfor Vision

Benchmark scores(2)

Scores are benchmark-specific and are direction-aware: the same numeric gap can mean very different outcomes across suites. Use the leaderboard context and this model's provider route to decide whether the winning margin is meaningful for your workload.
BenchmarkScoreVersionSource
MathVista75.6Thinking variant (accuracy%)https://huggingface.co/openbmb/MiniCPM-V-4.6
Massive Multi-discipline Multimodal Understanding55.3Thinking variant (accuracy%)https://huggingface.co/openbmb/MiniCPM-V-4.6

Migration checks

No linked migration route is available for this model yet.

Compare MiniCPM-V 4.6 with other models

Frequently asked questions

What is the context window of MiniCPM-V 4.6?

MiniCPM-V 4.6 has a context window of 262k tokens.

When was MiniCPM-V 4.6 released?

MiniCPM-V 4.6 was released on 2026-05-11.

What benchmarks has MiniCPM-V 4.6 been tested on?

MiniCPM-V 4.6 has been evaluated on 2 benchmarks, including MathVista, Massive Multi-discipline Multimodal Understanding.