MiniCPM-o 2.6
8B omni model with real-time speech-to-speech conversation
About
MiniCPM-o 2.6 is OpenBMB's 8-billion-parameter omni model, open-sourced on January 14, 2025 under Apache 2.0. It handles text, image, video and audio with real-time speech-to-speech conversation and multimodal live streaming, aimed at running on phones and edge devices. The team claims it beats GPT-4o mini and GPT-4V on single-image understanding, though these are team-reported figures. An int4 quantized variant runs in about 7GB of memory.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Team-reported: beats GPT-4o mini, GPT-4V and Gemini 1.5 Pro on single-image understanding (OpenCompass average 65.2); VideoMME 63.0 without subtitles.
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: OpenBMB with active repo and int4 variant; Apache 2.0 confirmed on HF card; Multimodal claims are team-reported; high HF download counts. License: Apache 2.0 (per HF model card)..
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 8.0/10
- Maintenance · recency, releases, issue load
- 8.0/10
- Openness · license + docs
- 8.0/10
- Manual review · sanity read + red-flag check
- 8.0/10
- · OpenBMB with active repo and int4 variant
- · Apache 2.0 confirmed on HF card
- · Multimodal claims are team-reported; high HF download counts