Model
Qwen2.5-VL-3B-Instruct
qwen2-5-vl-3b-instructRecord
- Family
- qwen
- Parameters
- 3.75
- HF downloads / 30d
- 3,907,819
Hugging Face model card
Identity
https://huggingface.co/Qwen/Qwen2.5-VL-3B-Instruct- Repository
- Qwen/Qwen2.5-VL-3B-Instruct
- Status
- known
- Link type
- Exact Hub repository
Public Hugging Face repository confirmed by the Hub API.
Leaderboard scores
Scraped public quality scores joined to this model by its Hugging Face repository. They describe the listed variant, not any specific local quantization, and are never speed measurements.
| Benchmark | Category | Score | Rank | Scored variant | Confidence |
|---|---|---|---|---|---|
| docvqa | multimodal | 96 | 8 | Qwen2.5-VL-3B-Instruct-AWQ | high |
| docvqa | multimodal | 91.6 | 52 | Qwen2.5-VL-3B-Instruct | medium |
| docvqa | multimodal | 91.6 | 54 | Qwen2.5-VL-3B-Instruct-unsloth-bnb-4bit | medium |
| docvqa | multimodal | 91.6 | 55 | Qwen2.5-VL-3B-Instruct-GGUF | medium |
| docvqa | multimodal | 91.6 | 56 | Qwen2.5-VL-3B-Instruct-GGUF | medium |
| docvqa | multimodal | 91.6 | 58 | Qwen2.5-VL-3B-Instruct-bnb-4bit | medium |
| docvqa | multimodal | 91.6 | 59 | Qwen2.5-VL-3B-Instruct | medium |
| mathvista | multimodal | 73.8 | 12 | Qwen2.5-VL-3B-Instruct-AWQ | high |
| mathvista | multimodal | 60.5 | 59 | Qwen2.5-VL-3B-Instruct | medium |
| mathvista | multimodal | 60.5 | 60 | Qwen2.5-VL-3B-Instruct-unsloth-bnb-4bit | medium |
| mathvista | multimodal | 60.5 | 61 | Qwen2.5-VL-3B-Instruct-GGUF | medium |
| mathvista | multimodal | 60.5 | 62 | Qwen2.5-VL-3B-Instruct-GGUF | medium |
| mathvista | multimodal | 60.5 | 63 | Qwen2.5-VL-3B-Instruct-bnb-4bit | medium |
| mathvista | multimodal | 60.5 | 64 | Qwen2.5-VL-3B-Instruct | medium |
| mmbench | multimodal | 87.9 | 3 | Qwen2.5-VL-3B-Instruct-AWQ | high |
| mmbench | multimodal | 79.3 | 52 | Qwen2.5-VL-3B-Instruct | medium |
| mmbench | multimodal | 79.3 | 53 | Qwen2.5-VL-3B-Instruct-unsloth-bnb-4bit | medium |
| mmbench | multimodal | 79.3 | 54 | Qwen2.5-VL-3B-Instruct-GGUF | medium |
| mmbench | multimodal | 79.3 | 55 | Qwen2.5-VL-3B-Instruct-GGUF | medium |
| mmbench | multimodal | 79.3 | 56 | Qwen2.5-VL-3B-Instruct-bnb-4bit | medium |
| mmbench | multimodal | 79.3 | 57 | Qwen2.5-VL-3B-Instruct | medium |
| mmbench | multimodal | 1.73 | 69 | Qwen2.5-VL-3B-Instruct | medium |
| mmbench | multimodal | 1.73 | 70 | Qwen2.5-VL-3B-Instruct-unsloth-bnb-4bit | medium |
| mmbench | multimodal | 1.73 | 71 | Qwen2.5-VL-3B-Instruct-GGUF | medium |
| mmbench | multimodal | 1.73 | 72 | Qwen2.5-VL-3B-Instruct-GGUF | medium |
| mmbench | multimodal | 1.73 | 73 | Qwen2.5-VL-3B-Instruct-bnb-4bit | medium |
| mmbench | multimodal | 1.73 | 74 | Qwen2.5-VL-3B-Instruct | medium |
| mmmu | multimodal | 52.3 | 165 | Qwen2.5-VL-3B-Instruct | medium |
| mmmu | multimodal | 52.3 | 166 | Qwen2.5-VL-3B-Instruct-unsloth-bnb-4bit | medium |
| mmmu | multimodal | 52.3 | 167 | Qwen2.5-VL-3B-Instruct-GGUF | medium |
| mmmu | multimodal | 52.3 | 168 | Qwen2.5-VL-3B-Instruct-GGUF | medium |
| mmmu | multimodal | 52.3 | 169 | Qwen2.5-VL-3B-Instruct-bnb-4bit | medium |
| mmmu | multimodal | 52.3 | 170 | Qwen2.5-VL-3B-Instruct | medium |
| mmmu | multimodal | 32.7 | 310 | Qwen2.5-VL-3B-Instruct | medium |
| mmmu | multimodal | 32.7 | 311 | Qwen2.5-VL-3B-Instruct-unsloth-bnb-4bit | medium |
| mmmu | multimodal | 32.7 | 312 | Qwen2.5-VL-3B-Instruct-GGUF | medium |
| mmmu | multimodal | 32.7 | 313 | Qwen2.5-VL-3B-Instruct-GGUF | medium |
| mmmu | multimodal | 32.7 | 314 | Qwen2.5-VL-3B-Instruct-bnb-4bit | medium |
| mmmu | multimodal | 32.7 | 315 | Qwen2.5-VL-3B-Instruct | medium |
Remaining fields
Identity, launch, related records, and measured speed are shown above. This is the rest of the normalized record.
- family
- qwen
- id
- qwen2-5-vl-3b-instruct
- name
- Qwen2.5-VL-3B-Instruct
- params
- 3.75
- schema version
- local-ai-registry/v1
downloads
- all time
- 100,478,431
- captured at
- 2026-08-31T15:38:41Z
- last 30d
- 3,907,819
Provenance & metadata (1)
source huggingface-api
Provenance & metadata (2)
facts
- active params · provenance · captured at
- 2026-08-31T13:13:53Z
active params · reason not-observed
active params · state unknown
- architecture · provenance · captured at
- 2026-08-31T13:13:53Z
architecture · reason not-observed
architecture · state unknown
- params · provenance · captured at
- 2026-08-31T13:13:53Z
params · reason safetensors-parameter-metadata
params · state known
provenance
- captured at
- 2026-08-31T13:13:53Z
sources
| captured at | kind | url |
|---|---|---|
| 2026-08-31T13:13:53Z | huggingface-api | huggingface.co/api/models/Qwen/Qwen2.5-VL-3B-Instruct ↗ |