GPT-5.4 vs DeepSeek-V3.2 (Thinking) Comparison

Comparing GPT-5.4 and DeepSeek-V3.2 (Thinking) across benchmarks, pricing, and capabilities.

Performance Benchmarks

Comparative analysis across standard metrics

5 benchmarks

GPT-5.4 outperforms in 5 benchmarks (BrowseComp, GPQA, Humanity's Last Exam, Terminal-Bench 2.0, Toolathlon), while DeepSeek-V3.2 (Thinking) is better at 0 benchmarks.

GPT-5.4 significantly outperforms across most benchmarks.

Tue Mar 17 2026 • llm-stats.com

Arena Performance

Human preference votes

Pricing Analysis

Price comparison per million tokens

DeepSeek-V3.2 (Thinking) costs less

For input processing, GPT-5.4 ($2.50/1M tokens) is 8.9x more expensive than DeepSeek-V3.2 (Thinking) ($0.28/1M tokens).

For output processing, GPT-5.4 ($15.00/1M tokens) is 35.7x more expensive than DeepSeek-V3.2 (Thinking) ($0.42/1M tokens).

In conclusion, GPT-5.4 is more expensive than DeepSeek-V3.2 (Thinking).*

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

Lowest available price from all providers
Tue Mar 17 2026 • llm-stats.com
OpenAI
GPT-5.4
Input tokens$2.50
Output tokens$15.00
Best providerOpenAI
DeepSeek
DeepSeek-V3.2 (Thinking)
Input tokens$0.28
Output tokens$0.42
Best providerDeepSeek
Notice missing or incorrect data?Start an Issue

Context Window

Maximum input and output token capacity

GPT-5.4 accepts 1,000,000 input tokens compared to DeepSeek-V3.2 (Thinking)'s 131,072 tokens. GPT-5.4 can generate longer responses up to 128,000 tokens, while DeepSeek-V3.2 (Thinking) is limited to 65,536 tokens.

OpenAI
GPT-5.4
Input1,000,000 tokens
Output128,000 tokens
DeepSeek
DeepSeek-V3.2 (Thinking)
Input131,072 tokens
Output65,536 tokens
Tue Mar 17 2026 • llm-stats.com

Input Capabilities

Supported data types and modalities

GPT-5.4 supports multimodal inputs, whereas DeepSeek-V3.2 (Thinking) does not.

GPT-5.4 can handle both text and other forms of data like images, making it suitable for multimodal applications.

GPT-5.4

Text
Images
Audio
Video

DeepSeek-V3.2 (Thinking)

Text
Images
Audio
Video

License

Usage and distribution terms

GPT-5.4 is licensed under a proprietary license, while DeepSeek-V3.2 (Thinking) uses MIT.

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

GPT-5.4

Proprietary

Closed source

DeepSeek-V3.2 (Thinking)

MIT

Open weights

Release Timeline

When each model was launched

GPT-5.4 was released on 2026-03-05, while DeepSeek-V3.2 (Thinking) was released on 2025-12-01.

GPT-5.4 is 3 months newer than DeepSeek-V3.2 (Thinking).

GPT-5.4

Mar 5, 2026

1 weeks ago

3mo newer
DeepSeek-V3.2 (Thinking)

Dec 1, 2025

3 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

GPT-5.4 is available from OpenAI. DeepSeek-V3.2 (Thinking) is available from DeepSeek. The availability of providers can affect quality of the model and reliability.

GPT-5.4

openai logo
OpenAI
Input Price:Input: $2.50/1MOutput Price:Output: $15.00/1M

DeepSeek-V3.2 (Thinking)

deepseek logo
DeepSeek
Input Price:Input: $0.28/1MOutput Price:Output: $0.42/1M
* Prices shown are per million tokens

Outputs Comparison

Notice missing or incorrect data?Start an Issue discussion

Key Takeaways

Larger context window (1,000,000 tokens)
Supports multimodal inputs
Higher BrowseComp score (82.7% vs 51.4%)
Higher GPQA score (92.8% vs 82.4%)
Higher Humanity's Last Exam score (39.8% vs 25.1%)
Higher Terminal-Bench 2.0 score (75.1% vs 46.4%)
Higher Toolathlon score (54.6% vs 35.2%)
Less expensive input tokens
Less expensive output tokens
Has open weights

Detailed Comparison

AI Model Comparison Table
Feature
OpenAI
GPT-5.4
DeepSeek
DeepSeek-V3.2 (Thinking)