Ovis2
Apache-2.0 open vision-language models (1B–34B) with strong small-model performance.
About
Alibaba's AIDC-AI team built Ovis2 around structural alignment of visual and textual embeddings, shipping sizes from 1B to 34B under Apache 2.0. The 2025 release added video and multi-image processing, multilingual OCR, and high-resolution image handling, with the small models punching above their size class.
It gets less English-language coverage than the big-name VLMs, so community tooling and docs are thinner. A newer Ovis2.5 exists if you want the latest checkpoint.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
- Leading results under 40B params on STEM, chart, grounding, and video benchmarks (team-reported)
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 6/10
Scored 6/10 (metrics tier, metrics-v1). Evidence: AIDC-AI (Alibaba) GitHub repo; Ovis2 launched 2025-01-26 in 1B-34B sizes; Apache-2.0 confirmed from repo license docs; Thinner English-language coverage and community tooling vs big-name VLMs; Benchmarks are team-reported. License: Apache-2.0 confirmed from Ovis repo license documentation..
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 6.0/10
- Maintenance · recency, releases, issue load
- 7.0/10
- Openness · license + docs
- 8.0/10
- Manual review · sanity read + red-flag check
- 6.0/10
- · AIDC-AI (Alibaba) GitHub repo; Ovis2 launched 2025-01-26 in 1B-34B sizes
- · Apache-2.0 confirmed from repo license docs
- · Thinner English-language coverage and community tooling vs big-name VLMs
- · Benchmarks are team-reported