Model Comparison

DeepSeek-R1-0528 vs Qwen3 VL 4B Thinking

DeepSeek-R1-0528 significantly outperforms across most benchmarks. Qwen3 VL 4B Thinking is 2.8x cheaper per token.

Want to compare interactively?Try the playground

Performance Benchmarks

Comparative analysis across standard metrics

4 benchmarks

DeepSeek-R1-0528 outperforms in 4 benchmarks (AIME 2025, GPQA, MMLU-Pro, MMLU-Redux), while Qwen3 VL 4B Thinking is better at 0 benchmarks.

DeepSeek-R1-0528 significantly outperforms across most benchmarks.

Sat May 16 2026 • llm-stats.com

Arena Performance

Human preference votes

Pricing Analysis

Price comparison per million tokens

Qwen3 VL 4B Thinking costs less

For input processing, DeepSeek-R1-0528 ($0.50/1M tokens) is 5.0x more expensive than Qwen3 VL 4B Thinking ($0.10/1M tokens).

For output processing, DeepSeek-R1-0528 ($2.15/1M tokens) is 2.1x more expensive than Qwen3 VL 4B Thinking ($1.00/1M tokens).

In conclusion, DeepSeek-R1-0528 is more expensive than Qwen3 VL 4B Thinking.*

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

Lowest available price from all providers

Sat May 16 2026 • llm-stats.com

DeepSeek-R1-0528

Input tokens$0.50

Output tokens$2.15

Best providerDeepinfra

Qwen3 VL 4B Thinking

Input tokens$0.10

Output tokens$1.00

Best providerDeepinfra

Notice missing or incorrect data?Start an Issue→

Model Size

Parameter count comparison

667.0B diff

DeepSeek-R1-0528 has 667.0B more parameters than Qwen3 VL 4B Thinking, making it 16675.0% larger.

DeepSeek-R1-0528

671.0Bparameters

Qwen3 VL 4B Thinking

4.0Bparameters

671.0B

DeepSeek-R1-0528

4.0B

Qwen3 VL 4B Thinking

Context Window

Maximum input and output token capacity

Qwen3 VL 4B Thinking accepts 262,144 input tokens compared to DeepSeek-R1-0528's 131,072 tokens. Qwen3 VL 4B Thinking can generate longer responses up to 262,144 tokens, while DeepSeek-R1-0528 is limited to 131,072 tokens.

DeepSeek-R1-0528

Input131,072 tokens

Output131,072 tokens

Qwen3 VL 4B Thinking

Input262,144 tokens

Output262,144 tokens

Sat May 16 2026 • llm-stats.com

Input Capabilities

Supported data types and modalities

Qwen3 VL 4B Thinking supports multimodal inputs, whereas DeepSeek-R1-0528 does not.

Qwen3 VL 4B Thinking can handle both text and other forms of data like images, making it suitable for multimodal applications.

DeepSeek-R1-0528

Text

Images

Audio

Video

Qwen3 VL 4B Thinking

Text

Images

Audio

Video

License

Usage and distribution terms

DeepSeek-R1-0528 is licensed under MIT, while Qwen3 VL 4B Thinking uses Apache 2.0.

License differences may affect how you can use these models in commercial or open-source projects.

DeepSeek-R1-0528

MIT

Open weights

Qwen3 VL 4B Thinking

Apache 2.0

Open weights

Release Timeline

When each model was launched

DeepSeek-R1-0528 was released on 2025-05-28, while Qwen3 VL 4B Thinking was released on 2025-09-22.

Qwen3 VL 4B Thinking is 4 months newer than DeepSeek-R1-0528.

DeepSeek-R1-0528

May 28, 2025

11 months ago

Qwen3 VL 4B Thinking

Sep 22, 2025

7 months ago

3mo newer

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

DeepSeek-R1-0528 is available from DeepInfra, DeepSeek, Novita. Qwen3 VL 4B Thinking is available from DeepInfra.

DeepSeek-R1-0528

Deepinfra

Input Price:Input: $0.50/1MOutput Price:Output: $2.15/1M

DeepSeek

Input Price:Input: $0.55/1MOutput Price:Output: $2.19/1M

Novita

Input Price:Input: $0.70/1MOutput Price:Output: $2.50/1M

Qwen3 VL 4B Thinking

Deepinfra

Input Price:Input: $0.10/1MOutput Price:Output: $1.00/1M

* Prices shown are per million tokens

Outputs Comparison

Notice missing or incorrect data?Start an Issue discussion→

Key Takeaways

DeepSeek-R1-0528

View details

DeepSeek

Higher AIME 2025 score (87.5% vs 74.5%)

Higher GPQA score (81.0% vs 64.1%)

Higher MMLU-Pro score (85.0% vs 73.6%)

Higher MMLU-Redux score (93.4% vs 86.0%)

Qwen3 VL 4B Thinking

View details

Alibaba Cloud / Qwen Team

Larger context window (262,144 tokens)

Supports multimodal inputs

Less expensive input tokens

Less expensive output tokens

Detailed Comparison

AI Model Comparison Table
Feature	DeepSeek-R1-0528	Qwen3 VL 4B Thinking

FAQ

Common questions about DeepSeek-R1-0528 vs Qwen3 VL 4B Thinking.

Which is better, DeepSeek-R1-0528 or Qwen3 VL 4B Thinking?

DeepSeek-R1-0528 significantly outperforms across most benchmarks. DeepSeek-R1-0528 is made by DeepSeek and Qwen3 VL 4B 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 DeepSeek-R1-0528 compare to Qwen3 VL 4B Thinking in benchmarks?

DeepSeek-R1-0528 scores MMLU-Redux: 93.4%, SimpleQA: 92.3%, AIME 2024: 91.4%, AIME 2025: 87.5%, MMLU-Pro: 85.0%. Qwen3 VL 4B Thinking scores DocVQAtest: 94.2%, ScreenSpot: 92.9%, MMBench-V1.1: 86.7%, MMLU-Redux: 86.0%, AI2D: 84.9%.

Is DeepSeek-R1-0528 cheaper than Qwen3 VL 4B Thinking?

Qwen3 VL 4B Thinking is 5.0x cheaper for input tokens. DeepSeek-R1-0528 costs $0.50/M input and $2.15/M output via deepinfra. Qwen3 VL 4B Thinking costs $0.10/M input and $1.00/M output via deepinfra.

What are the context window sizes for DeepSeek-R1-0528 and Qwen3 VL 4B Thinking?

DeepSeek-R1-0528 supports 131K tokens and Qwen3 VL 4B 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 DeepSeek-R1-0528 and Qwen3 VL 4B Thinking?

Key differences include context window (131K vs 262K), input pricing ($0.50 vs $0.10/M), multimodal support (no vs yes), licensing (MIT vs Apache 2.0). See the full comparison above for benchmark-by-benchmark results.

Who makes DeepSeek-R1-0528 and Qwen3 VL 4B Thinking?

DeepSeek-R1-0528 is developed by DeepSeek and Qwen3 VL 4B Thinking is developed by Alibaba Cloud / Qwen Team.