Model Comparison

Qwen3 VL 235B A22B Instruct vs Qwen3 VL 235B A22B ThinkingWhich is better in 2026?

Qwen3 VL 235B A22B Thinking has a slight edge in benchmark performance. Qwen3 VL 235B A22B Instruct is 2.0x cheaper per token.

Verdict: Qwen3 VL 235B A22B Instruct vs Qwen3 VL 235B A22B Thinking — which is better?

Qwen3 VL 235B A22B Instruct (by Alibaba Cloud / Qwen Team) and Qwen3 VL 235B A22B Thinking (by Alibaba Cloud / Qwen Team) are two of the AI models people compare most. Here is how they stack up on benchmarks, price and capabilities, and which one to pick in 2026.

Qwen3 VL 235B A22B Instruct outperforms in 15 benchmarks (AI2D, BLINK, CC-OCR, CharadesSTA, Creative Writing v3, DocVQAtest, LVBench, MLVU, OCRBench, OCRBench-V2 (en), ODinW, OSWorld, ScreenSpot Pro, SimpleQA, VideoMME w/o sub.), while Qwen3 VL 235B A22B Thinking is better at 27 benchmarks (AIME 2025, BFCL-v3, CharXiv-R, ERQA, Hallusion Bench, HMMT25, IFEval, InfoVQAtest, LiveBench 20241125, LiveCodeBench v6, MathVision, MathVista-Mini, MMBench-V1.1, MMLU, MMLU-Pro, MMLU-ProX, MMLU-Redux, MMMU-Pro, MMMUval, MMStar, MuirBench, Multi-IF, OCRBench-V2 (zh), RealWorldQA, SuperGPQA, VideoMMMU, WritingBench). Qwen3 VL 235B A22B Thinking has a slight edge in benchmark performance.

On price, Qwen3 VL 235B A22B Instruct is roughly 2.0x cheaper per token on a blended 3:1 input/output basis, which adds up quickly at production volume.

Choose Qwen3 VL 235B A22B Instruct if…

  • cost matters — it's about 2.0x cheaper per token

Choose Qwen3 VL 235B A22B Thinking if…

  • you want the strongest raw capability — it leads on 30 of 45 shared benchmarks
Want to compare interactively?Try the playground

Performance Benchmarks

Comparative analysis across standard metrics

45 benchmarks

Qwen3 VL 235B A22B Instruct outperforms in 15 benchmarks (AI2D, BLINK, CC-OCR, CharadesSTA, Creative Writing v3, DocVQAtest, LVBench, MLVU, OCRBench, OCRBench-V2 (en), ODinW, OSWorld, ScreenSpot Pro, SimpleQA, VideoMME w/o sub.), while Qwen3 VL 235B A22B Thinking is better at 27 benchmarks (AIME 2025, BFCL-v3, CharXiv-R, ERQA, Hallusion Bench, HMMT25, IFEval, InfoVQAtest, LiveBench 20241125, LiveCodeBench v6, MathVision, MathVista-Mini, MMBench-V1.1, MMLU, MMLU-Pro, MMLU-ProX, MMLU-Redux, MMMU-Pro, MMMUval, MMStar, MuirBench, Multi-IF, OCRBench-V2 (zh), RealWorldQA, SuperGPQA, VideoMMMU, WritingBench).

Qwen3 VL 235B A22B Thinking has a slight edge in benchmark performance.

Sun Jun 07 2026 • llm-stats.com

Arena Performance

Human preference votes

Pricing Analysis

Price comparison per million tokens

Qwen3 VL 235B A22B Instruct costs less

For input processing, Qwen3 VL 235B A22B Instruct ($0.30/1M tokens) is 1.5x cheaper than Qwen3 VL 235B A22B Thinking ($0.45/1M tokens).

For output processing, Qwen3 VL 235B A22B Instruct ($1.49/1M tokens) is 2.3x cheaper than Qwen3 VL 235B A22B Thinking ($3.49/1M tokens).

In conclusion, Qwen3 VL 235B A22B Thinking is more expensive than Qwen3 VL 235B A22B Instruct.*

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

Lowest available price from all providers
Sun Jun 07 2026 • llm-stats.com
Alibaba Cloud / Qwen Team
Qwen3 VL 235B A22B Instruct
Input tokens$0.30
Output tokens$1.49
Best providerDeepinfra
Alibaba Cloud / Qwen Team
Qwen3 VL 235B A22B Thinking
Input tokens$0.45
Output tokens$3.49
Best providerDeepinfra
Notice missing or incorrect data?Start an Issue→

Model Size

Parameter count comparison

0.0M diff

Qwen3 VL 235B A22B Thinking has 0.0B more parameters than Qwen3 VL 235B A22B Instruct, making it 0.0% larger.

Alibaba Cloud / Qwen Team
Qwen3 VL 235B A22B Instruct
236.0Bparameters
Alibaba Cloud / Qwen Team
Qwen3 VL 235B A22B Thinking
236.0Bparameters
236.0B
Qwen3 VL 235B A22B Instruct
236.0B
Qwen3 VL 235B A22B Thinking

Context Window

Maximum input and output token capacity

Both models have the same input context window of 262,144 tokens. Both models can generate responses up to 262,144 tokens.

Alibaba Cloud / Qwen Team
Qwen3 VL 235B A22B Instruct
Input262,144 tokens
Output262,144 tokens
Alibaba Cloud / Qwen Team
Qwen3 VL 235B A22B Thinking
Input262,144 tokens
Output262,144 tokens
Sun Jun 07 2026 • llm-stats.com

Input Capabilities

Supported data types and modalities

Both Qwen3 VL 235B A22B Instruct and Qwen3 VL 235B A22B Thinking support multimodal inputs.

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

Qwen3 VL 235B A22B Instruct

Text
Images
Audio
Video

Qwen3 VL 235B A22B Thinking

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 235B A22B Instruct

Apache 2.0

Open weights

Qwen3 VL 235B A22B Thinking

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 235B A22B Instruct

Sep 22, 2025

8 months ago

Qwen3 VL 235B A22B Thinking

Sep 22, 2025

8 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 235B A22B Instruct is available from DeepInfra, Novita. Qwen3 VL 235B A22B Thinking is available from DeepInfra, Novita.

Qwen3 VL 235B A22B Instruct

deepinfra logo
Deepinfra
Input Price:Input: $0.30/1MOutput Price:Output: $1.49/1M
novita logo
Novita
Input Price:Input: $0.30/1MOutput Price:Output: $1.50/1M

Qwen3 VL 235B A22B Thinking

deepinfra logo
Deepinfra
Input Price:Input: $0.45/1MOutput Price:Output: $3.49/1M
novita logo
Novita
Input Price:Input: $0.98/1MOutput Price:Output: $3.95/1M
* Prices shown are per million tokens

Outputs Comparison

Notice missing or incorrect data?Start an Issue discussion→

Key Takeaways

Alibaba Cloud / Qwen Team
Less expensive input tokens
Less expensive output tokens
Higher AI2D score (89.7% vs 89.2%)
Higher BLINK score (70.7% vs 67.1%)
Higher CC-OCR score (82.2% vs 81.5%)
Higher CharadesSTA score (64.8% vs 63.5%)
Higher Creative Writing v3 score (86.5% vs 85.7%)
Higher DocVQAtest score (97.1% vs 96.5%)
Higher LVBench score (67.7% vs 63.6%)
Higher MLVU score (84.3% vs 83.8%)
Higher OCRBench score (92.0% vs 87.5%)
Higher OCRBench-V2 (en) score (67.1% vs 66.8%)
Higher ODinW score (48.6% vs 43.2%)
Higher OSWorld score (66.7% vs 38.1%)
Higher ScreenSpot Pro score (62.0% vs 61.8%)
Higher SimpleQA score (51.9% vs 44.4%)
Higher VideoMME w/o sub. score (79.2% vs 79.0%)
Alibaba Cloud / Qwen Team
Higher AIME 2025 score (89.7% vs 74.7%)
Higher BFCL-v3 score (71.9% vs 67.7%)
Higher CharXiv-R score (66.1% vs 62.1%)
Higher ERQA score (52.5% vs 51.3%)
Higher Hallusion Bench score (66.7% vs 63.2%)
Higher HMMT25 score (77.4% vs 57.4%)
Higher IFEval score (88.2% vs 87.8%)
Higher InfoVQAtest score (89.5% vs 89.2%)
Higher LiveBench 20241125 score (79.6% vs 74.8%)
Higher LiveCodeBench v6 score (70.1% vs 54.3%)
Higher MathVision score (74.6% vs 66.5%)
Higher MathVista-Mini score (85.8% vs 84.9%)
Higher MMBench-V1.1 score (90.6% vs 89.9%)
Higher MMLU score (90.6% vs 88.8%)
Higher MMLU-Pro score (83.8% vs 81.8%)
Higher MMLU-ProX score (80.6% vs 77.8%)
Higher MMLU-Redux score (93.7% vs 92.2%)
Higher MMMU-Pro score (69.3% vs 68.1%)
Higher MMMUval score (80.6% vs 78.7%)
Higher MMStar score (78.7% vs 78.4%)
Higher MuirBench score (80.1% vs 72.8%)
Higher Multi-IF score (79.1% vs 76.3%)
Higher OCRBench-V2 (zh) score (63.5% vs 61.8%)
Higher RealWorldQA score (81.3% vs 79.3%)
Higher SuperGPQA score (64.3% vs 60.4%)
Higher VideoMMMU score (80.0% vs 74.7%)
Higher WritingBench score (86.7% vs 85.5%)

Detailed Comparison

AI Model Comparison Table
Feature
Alibaba Cloud / Qwen Team
Qwen3 VL 235B A22B Instruct
Alibaba Cloud / Qwen Team
Qwen3 VL 235B A22B Thinking

FAQ

Common questions about Qwen3 VL 235B A22B Instruct vs Qwen3 VL 235B A22B Thinking.

Which is better, Qwen3 VL 235B A22B Instruct or Qwen3 VL 235B A22B Thinking?

Qwen3 VL 235B A22B Thinking has a slight edge in benchmark performance. Qwen3 VL 235B A22B Instruct is made by Alibaba Cloud / Qwen Team and Qwen3 VL 235B A22B Thinking 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 235B A22B Instruct compare to Qwen3 VL 235B A22B Thinking in benchmarks?

Qwen3 VL 235B A22B Instruct scores DocVQAtest: 97.1%, ScreenSpot: 95.4%, MMLU-Redux: 92.2%, OCRBench: 92.0%, MMBench-V1.1: 89.9%. Qwen3 VL 235B A22B Thinking scores ZebraLogic: 97.3%, DocVQAtest: 96.5%, ScreenSpot: 95.4%, CountBench: 93.7%, MMLU-Redux: 93.7%.

Is Qwen3 VL 235B A22B Instruct cheaper than Qwen3 VL 235B A22B Thinking?

Qwen3 VL 235B A22B Instruct is 1.5x cheaper for input tokens. Qwen3 VL 235B A22B Instruct costs $0.30/M input and $1.49/M output via deepinfra. Qwen3 VL 235B A22B Thinking costs $0.45/M input and $3.49/M output via deepinfra.

What are the context window sizes for Qwen3 VL 235B A22B Instruct and Qwen3 VL 235B A22B Thinking?

Qwen3 VL 235B A22B Instruct supports 262K tokens and Qwen3 VL 235B A22B Thinking supports 262K 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 235B A22B Instruct and Qwen3 VL 235B A22B Thinking?

Key differences include input pricing ($0.30 vs $0.45/M). See the full comparison above for benchmark-by-benchmark results.

Related comparisons