Model Comparison

Qwen3 VL 30B A3B Instruct vs Qwen3 VL 8B Instruct

Qwen3 VL 30B A3B Instruct shows notably better performance in the majority of benchmarks. Qwen3 VL 8B Instruct is 1.8x cheaper per token.

Want to compare interactively?Try the playground

Performance Benchmarks

Comparative analysis across standard metrics

46 benchmarks

Qwen3 VL 30B A3B Instruct outperforms in 34 benchmarks (AIME 2025, CC-OCR, CharadesSTA, CharXiv-D, CharXiv-R, Hallusion Bench, HMMT25, IFEval, Include, LiveBench 20241125, LiveCodeBench v6, LVBench, MathVision, MathVista-Mini, MLVU-M, MMBench-V1.1, MMLU, MMLU-Pro, MMLU-ProX, MMLU-Redux, MM-MT-Bench, MMMU-Pro, MMMU (val), MMStar, MVBench, OCRBench, ODinW, PolyMATH, RealWorldQA, ScreenSpot, ScreenSpot Pro, SuperGPQA, Video-MME, VideoMMMU), while Qwen3 VL 8B Instruct is better at 11 benchmarks (AI2D, BLINK, DocVQAtest, ERQA, InfoVQAtest, MuirBench, Multi-IF, OCRBench-V2 (en), OCRBench-V2 (zh), OSWorld, WritingBench).

Qwen3 VL 30B A3B Instruct shows notably better performance in the majority of benchmarks.

Fri May 08 2026 • llm-stats.com

Arena Performance

Human preference votes

Pricing Analysis

Price comparison per million tokens

Qwen3 VL 8B Instruct costs less

For input processing, Qwen3 VL 30B A3B Instruct ($0.20/1M tokens) is 2.5x more expensive than Qwen3 VL 8B Instruct ($0.08/1M tokens).

For output processing, Qwen3 VL 30B A3B Instruct ($0.70/1M tokens) is 1.4x more expensive than Qwen3 VL 8B Instruct ($0.50/1M tokens).

In conclusion, Qwen3 VL 30B A3B Instruct is more expensive than Qwen3 VL 8B Instruct.*

* Using a 3:1 ratio of input to output tokens

Lowest available price from all providers

Fri May 08 2026 • llm-stats.com

Qwen3 VL 30B A3B Instruct

Input tokens$0.20

Output tokens$0.70

Best providerNovita

Qwen3 VL 8B Instruct

Input tokens$0.08

Output tokens$0.50

Best providerNovita

Notice missing or incorrect data?Start an Issue→

Model Size

Parameter count comparison

22.0B diff

Qwen3 VL 30B A3B Instruct has 22.0B more parameters than Qwen3 VL 8B Instruct, making it 244.4% larger.

Qwen3 VL 30B A3B Instruct

31.0Bparameters

Qwen3 VL 8B Instruct

9.0Bparameters

31.0B

Qwen3 VL 30B A3B Instruct

9.0B

Qwen3 VL 8B Instruct

Context Window

Maximum input and output token capacity

Both models have the same input context window of 131,072 tokens. Both models can generate responses up to 32,768 tokens.

Qwen3 VL 30B A3B Instruct

Input131,072 tokens

Output32,768 tokens

Qwen3 VL 8B Instruct

Input131,072 tokens

Output32,768 tokens

Fri May 08 2026 • llm-stats.com

Input Capabilities

Supported data types and modalities

Both Qwen3 VL 30B A3B Instruct and Qwen3 VL 8B Instruct support multimodal inputs.

They are both capable of processing various types of data, offering versatility in application.

Qwen3 VL 30B A3B Instruct

Text

Images

Audio

Video

Qwen3 VL 8B Instruct

Text

Images

Audio

Video

License

Usage and distribution terms

Both models are licensed under Apache 2.0.

Both models share the same licensing terms, providing consistent usage rights.

Qwen3 VL 30B A3B Instruct

Apache 2.0

Open weights

Qwen3 VL 8B Instruct

Apache 2.0

Open weights

Release Timeline

When each model was launched

Both models were released on 2025-09-22.

They likely represent similar generations of model development.

Qwen3 VL 30B A3B Instruct

Sep 22, 2025

7 months ago

Qwen3 VL 8B Instruct

Sep 22, 2025

7 months ago

Knowledge Cutoff

When training data ends

Neither model specifies a knowledge cutoff date.

Unable to compare the recency of their training data.

No cutoff dates available

Provider Availability

Qwen3 VL 30B A3B Instruct is available from Novita, DeepInfra. Qwen3 VL 8B Instruct is available from Novita, DeepInfra.

Qwen3 VL 30B A3B Instruct

Novita

Input Price:Input: $0.20/1MOutput Price:Output: $0.70/1M

Deepinfra

Input Price:Input: $0.29/1MOutput Price:Output: $0.99/1M

Qwen3 VL 8B Instruct

Novita

Input Price:Input: $0.08/1MOutput Price:Output: $0.50/1M

Deepinfra

Input Price:Input: $0.18/1MOutput Price:Output: $0.69/1M

* Prices shown are per million tokens

Outputs Comparison

Notice missing or incorrect data?Start an Issue discussion→

Key Takeaways

Qwen3 VL 30B A3B Instruct

View details

Alibaba Cloud / Qwen Team

Higher AIME 2025 score (69.3% vs 45.9%)

Higher CC-OCR score (80.7% vs 79.9%)

Higher CharadesSTA score (63.5% vs 56.0%)

Higher CharXiv-D score (85.5% vs 83.0%)

Higher CharXiv-R score (48.9% vs 46.4%)

Higher Hallusion Bench score (61.5% vs 61.1%)

Higher HMMT25 score (50.6% vs 32.5%)

Higher IFEval score (85.8% vs 83.7%)

Higher Include score (71.6% vs 67.0%)

Higher LiveBench 20241125 score (65.4% vs 62.0%)

Higher LiveCodeBench v6 score (42.6% vs 39.3%)

Higher LVBench score (62.5% vs 58.0%)

Higher MathVision score (60.2% vs 53.9%)

Higher MathVista-Mini score (80.1% vs 77.2%)

Higher MLVU-M score (81.3% vs 78.1%)

Higher MMBench-V1.1 score (87.0% vs 85.0%)

Higher MMLU score (85.0% vs 80.7%)

Higher MMLU-Pro score (77.8% vs 71.6%)

Higher MMLU-ProX score (70.9% vs 65.4%)

Higher MMLU-Redux score (88.4% vs 84.9%)

Higher MM-MT-Bench score (8.1% vs 7.7%)

Higher MMMU-Pro score (60.4% vs 55.9%)

Higher MMMU (val) score (74.2% vs 69.6%)

Higher MMStar score (72.1% vs 70.9%)

Higher MVBench score (72.3% vs 68.7%)

Higher OCRBench score (90.3% vs 89.6%)

Higher ODinW score (47.5% vs 44.7%)

Higher PolyMATH score (44.3% vs 30.4%)

Higher RealWorldQA score (73.7% vs 71.5%)

Higher ScreenSpot score (94.7% vs 94.4%)

Higher ScreenSpot Pro score (60.5% vs 54.6%)

Higher SuperGPQA score (53.1% vs 44.5%)

Higher Video-MME score (74.5% vs 71.4%)

Higher VideoMMMU score (68.7% vs 65.3%)

Qwen3 VL 8B Instruct

View details

Alibaba Cloud / Qwen Team

Less expensive input tokens

Less expensive output tokens

Higher AI2D score (85.7% vs 85.0%)

Higher BLINK score (69.1% vs 67.7%)

Higher DocVQAtest score (96.1% vs 95.0%)

Higher ERQA score (45.8% vs 43.0%)

Higher InfoVQAtest score (83.1% vs 82.0%)

Higher MuirBench score (64.4% vs 62.9%)

Higher Multi-IF score (75.1% vs 66.1%)

Higher OCRBench-V2 (en) score (65.4% vs 63.2%)

Higher OCRBench-V2 (zh) score (61.2% vs 57.8%)

Higher OSWorld score (33.9% vs 30.3%)

Higher WritingBench score (83.1% vs 82.6%)

Detailed Comparison

AI Model Comparison Table
Feature	Qwen3 VL 30B A3B Instruct	Qwen3 VL 8B Instruct

FAQ

Common questions about Qwen3 VL 30B A3B Instruct vs Qwen3 VL 8B Instruct.

Which is better, Qwen3 VL 30B A3B Instruct or Qwen3 VL 8B Instruct?

Qwen3 VL 30B A3B Instruct shows notably better performance in the majority of benchmarks. Qwen3 VL 30B A3B Instruct is made by Alibaba Cloud / Qwen Team and Qwen3 VL 8B Instruct is made by Alibaba Cloud / Qwen Team. The best choice depends on your use case — compare their benchmark scores, pricing, and capabilities above.

How does Qwen3 VL 30B A3B Instruct compare to Qwen3 VL 8B Instruct in benchmarks?

Qwen3 VL 30B A3B Instruct scores DocVQAtest: 95.0%, ScreenSpot: 94.7%, OCRBench: 90.3%, MMLU-Redux: 88.4%, MMBench-V1.1: 87.0%. Qwen3 VL 8B Instruct scores DocVQAtest: 96.1%, ScreenSpot: 94.4%, OCRBench: 89.6%, AI2D: 85.7%, MMBench-V1.1: 85.0%.

Is Qwen3 VL 30B A3B Instruct cheaper than Qwen3 VL 8B Instruct?

Qwen3 VL 8B Instruct is 2.5x cheaper for input tokens. Qwen3 VL 30B A3B Instruct costs $0.20/M input and $0.70/M output via novita. Qwen3 VL 8B Instruct costs $0.08/M input and $0.50/M output via novita.

What are the context window sizes for Qwen3 VL 30B A3B Instruct and Qwen3 VL 8B Instruct?

Qwen3 VL 30B A3B Instruct supports 131K tokens and Qwen3 VL 8B Instruct supports 131K tokens. A larger context window lets you process longer documents, conversations, or codebases in a single request.

What are the main differences between Qwen3 VL 30B A3B Instruct and Qwen3 VL 8B Instruct?

Key differences include input pricing ($0.20 vs $0.08/M). See the full comparison above for benchmark-by-benchmark results.