QVQ 72B Preview
Dec 2024
QVQ 72B PreviewCompare inonMMMU validation
Alternatives?
ModelScoreCostPer taskWhat changes
CheaperLlama 4 ScoutunstatedOpensame-52%$0.10 / $0.30Memory (4-bit): 43 GB → 65 GB
BetterGPT-5.1unstated+21%+1,000%$1.25 / $10
OpenSkywork R1V3 38B+8%——
CheaperGemma 3 27BunstatedOpen-8%-68%$0.08 / $0.16Bettero3unstated+18%+1,020%$2 / $8ModelScoreCostPer taskWhat changes
Llama 4 ScoutunstatedOpensame-52%$0.10 / $0.30Memory (4-bit): 43 GB → 65 GBGemma 3 27BunstatedOpen-8%-68%$0.08 / $0.16ModelScoreCostPer taskWhat changes
GPT-5.1unstated+21%+1,000%$1.25 / $10o3unstated+18%+1,020%$2 / $8
Claude Opus 4.5unstated+15%+3,100%$5 / $25dDots.vlm1.instOpen+14%——Memory (4-bit): not verifiedGemini 2.5 Pro Preview (May' 25)unstated+13%——No image inputModelScoreCostPer taskWhat changes
Skywork R1V3 38BOpen+8%——Skywork R1V2 38BOpen+5%——Llama 4 ScoutunstatedOpensame-52%$0.10 / $0.30Memory (4-bit): 43 GB → 65 GBSkywork R1V 38BOpen-2%——
MiMo VL 7B RLOpen-5%——ModelScoreCostPer taskWhat changes
Qwen3 VL 235B A22B+12%—$0.53 / $2.66Qwen3.8 Omni Flashn/a—$0.15 / $0.47Qwen3.8 Max (0902)thinkingn/a—$5.41Qwen3.8 FlashthinkingOpenn/a—$0.37Memory (4-bit): 43 GB → 108 GB+6Qwen3.8 27BmediumOpenn/a—$1.13max_tokens range · QwenCloud+4Qwen3.8 2.4T A95BthinkingOpenn/a—$2.16Memory (4-bit): 43 GB → 1,468 GB+6Qwen3.7 Flashunstatedn/a—$0.03 / $0.13Qwen3.7 Plusthinkingn/a—$0.22All Alibaba (Qwen) models →Run it yourself
estimateFits on✗ 8 GB (phone, laptop)✗ 16 GB✗ 24 GB GPU✓ 48 GB GPU · compressed✓ 80 GB GPU · 8-bit
Licenceqwencommercial use with conditions
Benchmarks
in Vision & DocumentsMMMU · validation70.3#14
FAQ
How much does QVQ 72B Preview cost?›
$0.25 per 1M input tokens and $0.50 per 1M output tokens (cheapest host: DeepInfra).
Is QVQ 72B Preview free to use?›
There is no free API; the cheapest paid price is $0.25 per 1M input tokens. The weights are open, so you can run it yourself.
How good is QVQ 72B Preview, and what is it best at?›
#14 of 41 in Vision & Documents.
Is QVQ 72B Preview better than GPT-4o?›
On MMMU · validation, QVQ 72B Preview scores 70.3 vs 70.7 — about the same.
How can I use QVQ 72B Preview?›
Through the API of DeepInfra.
What is the context window of QVQ 72B Preview?›
32K tokens.
When was QVQ 72B Preview released? Is there a newer version?›
Released 2024-12-24. It is the current version in its line.
Is QVQ 72B Preview open source?›
Yes, the weights are open (qwen licence).
Can I run QVQ 72B Preview locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
Can QVQ 72B Preview read images?›
It accepts text and images as input.
What is QVQ 72B Preview?›
QVQ 72B Preview is an open-weights language model from Alibaba (Qwen), released Dec 2024.
What are the best alternatives to QVQ 72B Preview?›
Llama 4 Scout (cheaper), GPT-5.1 (better), Skywork R1V3 38B (open weights).