Qianfan VL 3B
Sep 2025
Qianfan VL 3BCompare inonMMMU validation
—per task?
self-host onlyprice
—output speed?
33Kcontext
Alternatives?
ModelScoreCostPer taskWhat changes
BetterGPT-5.1unstated+84%—$1.25 / $10
OpenMiMo VL 7B RL+44%——Memory (4-bit): 2.2 GB → 4.9 GBBettero3unstated+79%—$2 / $8
BetterClaude Opus 4.5unstated+74%—$5 / $25dBetterDots.vlm1.instOpen+73%——Memory (4-bit): not verifiedModelScoreCostPer taskWhat changes
Nothing here yet.
ModelScoreCostPer taskWhat changes
GPT-5.1unstated+84%—$1.25 / $10o3unstated+79%—$2 / $8Claude Opus 4.5unstated+74%—$5 / $25dDots.vlm1.instOpen+73%——Memory (4-bit): not verified
Gemini 2.5 Pro Preview (May' 25)unstated+72%——No image inputModelScoreCostPer taskWhat changes
MiMo VL 7B RLOpen+44%——Memory (4-bit): 2.2 GB → 4.9 GBMiMo VL 7B SFTOpen+39%——Memory (4-bit): 2.2 GB → 4.9 GBOInternVL3 8BOpen+35%——Memory (4-bit): 2.2 GB → 4.8 GBGemma 3 12BunstatedOpen+28%—$0.05 / $0.10Memory (4-bit): 2.2 GB → 6.1 GB
Eagle2.5 8BOpen+20%——Context: not verified+1ModelScoreCostPer taskWhat changes
Qianfan VL 70BOpen+26%——Memory (4-bit): 2.2 GB → 42 GBQianfan VL 8BOpensame——Memory (4-bit): 2.2 GB → 5.2 GBUnlimited OCROpenn/a——Qianfan OCROpenn/a——Memory (4-bit): 2.2 GB → 2.8 GBERNIE 5.0 Thinking Previewn/a——PaddleOCR VLn/a—$0.02 / $0.02ERNIE 5.0n/a——Context: not verifiedAll Baidu models →Run it yourself
estimateFits on✓ 8 GB (phone, laptop) · full quality✓ 16 GB · full quality✓ 24 GB GPU · full quality✓ 48 GB GPU · full quality✓ 80 GB GPU · full quality
Licencecustom licencecommercial use with conditions
Benchmarks
in Vision & DocumentsMMMU · validation46.4#35
FAQ
How much does Qianfan VL 3B cost?›
There is no per-unit API price: self-host only.
Is Qianfan VL 3B free to use?›
There is no free API. The weights are open, so you can run it yourself.
How good is Qianfan VL 3B, and what is it best at?›
#35 of 41 in Vision & Documents.
Is Qianfan VL 3B better than Qianfan VL 8B?›
On MMMU · validation, Qianfan VL 3B scores 46.4 vs 47.1 — about the same.
What is the context window of Qianfan VL 3B?›
33K tokens.
When was Qianfan VL 3B released? Is there a newer version?›
Released 2025-09-23. It is the current version in its line.
Is Qianfan VL 3B open source?›
Yes, the weights are open.
Can I run Qianfan VL 3B locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
Can Qianfan VL 3B read images?›
It accepts text and images as input.
What is Qianfan VL 3B?›
Qianfan VL 3B is an open-weights language model from Baidu, released Sep 2025.
What are the best alternatives to Qianfan VL 3B?›
GPT-5.1 (better), MiMo VL 7B RL (open weights).