Model Comparison

GPT-4o vs Kimi K2-Instruct-0905

Kimi K2-Instruct-0905 significantly outperforms across most benchmarks.

Performance Benchmarks

Comparative analysis across standard metrics

13 benchmarks

GPT-4o outperforms in 2 benchmarks (Humanity's Last Exam, SimpleQA), while Kimi K2-Instruct-0905 is better at 11 benchmarks (Aider-Polyglot, AIME 2024, GPQA, IFEval, MMLU, MMLU-Pro, Multi-Challenge, SWE-Bench Verified, Tau2 Airline, Tau2 Retail, Tau2 Telecom).

Kimi K2-Instruct-0905 significantly outperforms across most benchmarks.

Mon May 25 2026 • llm-stats.com

Arena Performance

Human preference votes

Context Window

Maximum input and output token capacity

Only GPT-4o specifies input context (128,000 tokens). Only GPT-4o specifies output context (16,384 tokens).

OpenAI
GPT-4o
Input128,000 tokens
Output16,384 tokens
Moonshot AI
Kimi K2-Instruct-0905
Input- tokens
Output- tokens
Mon May 25 2026 • llm-stats.com

Input Capabilities

Supported data types and modalities

GPT-4o supports multimodal inputs, whereas Kimi K2-Instruct-0905 does not.

GPT-4o can handle both text and other forms of data like images, making it suitable for multimodal applications.

GPT-4o

Text
Images
Audio
Video

Kimi K2-Instruct-0905

Text
Images
Audio
Video

License

Usage and distribution terms

GPT-4o is licensed under a proprietary license, while Kimi K2-Instruct-0905 uses MIT.

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

GPT-4o

Proprietary

Closed source

Kimi K2-Instruct-0905

MIT

Open weights

Release Timeline

When each model was launched

GPT-4o was released on 2024-08-06, while Kimi K2-Instruct-0905 was released on 2025-09-05.

Kimi K2-Instruct-0905 is 13 months newer than GPT-4o.

GPT-4o

Aug 6, 2024

1.8 years ago

Kimi K2-Instruct-0905

Sep 5, 2025

8 months ago

1.1yr 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

Outputs Comparison

Notice missing or incorrect data?Start an Issue discussion

Key Takeaways

Larger context window (128,000 tokens)
Supports multimodal inputs
Higher Humanity's Last Exam score (5.3% vs 4.7%)
Higher SimpleQA score (38.2% vs 31.0%)
Has open weights
Higher Aider-Polyglot score (60.0% vs 30.7%)
Higher AIME 2024 score (69.6% vs 13.1%)
Higher GPQA score (75.1% vs 70.1%)
Higher IFEval score (89.8% vs 81.0%)
Higher MMLU score (89.5% vs 85.7%)
Higher MMLU-Pro score (81.1% vs 74.7%)
Higher Multi-Challenge score (54.1% vs 40.3%)
Higher SWE-Bench Verified score (65.8% vs 33.2%)
Higher Tau2 Airline score (56.5% vs 45.5%)
Higher Tau2 Retail score (70.6% vs 63.4%)
Higher Tau2 Telecom score (65.8% vs 23.5%)

Detailed Comparison

AI Model Comparison Table
Feature
OpenAI
GPT-4o
Moonshot AI
Kimi K2-Instruct-0905

FAQ

Common questions about GPT-4o vs Kimi K2-Instruct-0905.

Which is better, GPT-4o or Kimi K2-Instruct-0905?

Kimi K2-Instruct-0905 significantly outperforms across most benchmarks. GPT-4o is made by OpenAI and Kimi K2-Instruct-0905 is made by Moonshot AI. The best choice depends on your use case — compare their benchmark scores, pricing, and capabilities above.

How does GPT-4o compare to Kimi K2-Instruct-0905 in benchmarks?

GPT-4o scores AI2D: 94.2%, DocVQA: 92.8%, ChartQA: 85.7%, MMLU: 85.7%, CharXiv-D: 85.3%. Kimi K2-Instruct-0905 scores MATH-500: 97.4%, MMLU-Redux: 92.7%, IFEval: 89.8%, AutoLogi: 89.5%, MMLU: 89.5%.

What are the context window sizes for GPT-4o and Kimi K2-Instruct-0905?

GPT-4o supports 128K tokens and Kimi K2-Instruct-0905 supports an unknown number of tokens. A larger context window lets you process longer documents, conversations, or codebases in a single request.

What are the main differences between GPT-4o and Kimi K2-Instruct-0905?

Key differences include multimodal support (yes vs no), licensing (Proprietary vs MIT). See the full comparison above for benchmark-by-benchmark results.

Who makes GPT-4o and Kimi K2-Instruct-0905?

GPT-4o is developed by OpenAI and Kimi K2-Instruct-0905 is developed by Moonshot AI.