LLaVA-1.5-7B
The open-source vision-language baseline that kicked off the VLM era.
About
LLaVA-1.5-7B is the classic open vision-language model: a CLIP ViT-L-336px visual encoder connected to a Vicuna-1.5-7B language model through a simple MLP projection, fine-tuned on multimodal instruction-following data (558K pretraining image-text pairs, ~665K instruction samples including academic VQA data). Its minimalist architecture made state-of-the-art multimodal research cheap and reproducible in 2023, and it remains the standard baseline for VLM research.
Benchmarks below are author-reported from the LLaVA-1.5 paper (7B checkpoint): GQA 62.0, VisWiz 50.0, ScienceQA-IMG 66.8, VQA-v2 78.5, TextVQA 58.2, POPE 85.9, MME 1510.7, MMBench 64.3, MMBench-CN 58.3, SEED-Image 66.1, MM-Vet 30.5. Context window is 4,096 tokens (verified from the HF config.json max_position_embeddings). Weights fall under the LLaMA 2 Community License — not fully open for all commercial uses.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Author-reported (LLaVA-1.5 paper, 7B checkpoint):
- GQA: 62.0 | VisWiz: 50.0 | ScienceQA-IMG: 66.8
- VQA-v2: 78.5 | TextVQA: 58.2 | POPE: 85.9
- MME: 1510.7 | MMBench: 64.3 | MMBench-CN: 58.3
- SEED-Image: 66.1 | MM-Vet: 30.5
Third-party reproductions (MMBench test ~66.5-67.7) roughly corroborate.
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 7/10
Scored 7/10 (verified tier, metrics-v1). Evidence: HF: liuhaotian/llava-v1.5-7b; HF downloads 83,829 (peer-normalized); released 1104 days ago. Manual review: context 4096 verified from HF config.json; benchmarks author-reported; third-party MMBench reproductions corroborate. License: LLaMA 2 Community License (not OSI).
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 7.6/10
- Maintenance · recency, releases, issue load
- 4.0/10
- Openness · license + docs
- 7.2/10
- Manual review · sanity read + red-flag check
- 7.0/10
- · HF: liuhaotian/llava-v1.5-7b
- · HF downloads 83,829 (peer-normalized)
- · released 1104 days ago
- · manual review: context 4096 verified from HF config.json; benchmarks author-reported; third-party MMBench reproductions corroborate