Model Comparison

Gemini 2.5 Flash-Lite vs Kimi K2-Instruct-0905Which is better in 2026?

Kimi K2-Instruct-0905 shows notably better performance in the majority of benchmarks.

Verdict: Gemini 2.5 Flash-Lite vs Kimi K2-Instruct-0905 — which is better?

Gemini 2.5 Flash-Lite (by Google) and Kimi K2-Instruct-0905 (by Moonshot AI) 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.

Gemini 2.5 Flash-Lite outperforms in 2 benchmarks (AIME 2025, Humanity's Last Exam), while Kimi K2-Instruct-0905 is better at 5 benchmarks (Aider-Polyglot, GPQA, LiveCodeBench, SimpleQA, SWE-Bench Verified). Kimi K2-Instruct-0905 shows notably better performance in the majority of benchmarks.

Choose Gemini 2.5 Flash-Lite if…

  • you want predictable pricing at $0.10/M input and $0.40/M output

Choose Kimi K2-Instruct-0905 if…

  • you want the strongest raw capability — it leads on 5 of 7 shared benchmarks
  • you want the most recent training data — it shipped Sep 2025

Performance Benchmarks

Comparative analysis across standard metrics

7 benchmarks

Gemini 2.5 Flash-Lite outperforms in 2 benchmarks (AIME 2025, Humanity's Last Exam), while Kimi K2-Instruct-0905 is better at 5 benchmarks (Aider-Polyglot, GPQA, LiveCodeBench, SimpleQA, SWE-Bench Verified).

Kimi K2-Instruct-0905 shows notably better performance in the majority of benchmarks.

Tue Jul 28 2026 • llm-stats.com

Arena Performance

Human preference votes

Context Window

Maximum input and output token capacity

Only Gemini 2.5 Flash-Lite specifies input context (1,048,576 tokens). Only Gemini 2.5 Flash-Lite specifies output context (65,536 tokens).

Google
Gemini 2.5 Flash-Lite
Input1,048,576 tokens
Output65,536 tokens
Moonshot AI
Kimi K2-Instruct-0905
Input- tokens
Output- tokens
Tue Jul 28 2026 • llm-stats.com

Input Capabilities

Supported data types and modalities

Gemini 2.5 Flash-Lite supports multimodal inputs, whereas Kimi K2-Instruct-0905 does not.

Gemini 2.5 Flash-Lite can handle both text and other forms of data like images, making it suitable for multimodal applications.

Gemini 2.5 Flash-Lite

Text
Images
Audio
Video

Kimi K2-Instruct-0905

Text
Images
Audio
Video

License

Usage and distribution terms

Gemini 2.5 Flash-Lite is licensed under Creative Commons Attribution 4.0 License, while Kimi K2-Instruct-0905 uses MIT.

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

Gemini 2.5 Flash-Lite

Creative Commons Attribution 4.0 License

Open weights

Kimi K2-Instruct-0905

MIT

Open weights

Release Timeline

When each model was launched

Gemini 2.5 Flash-Lite was released on 2025-06-17, while Kimi K2-Instruct-0905 was released on 2025-09-05.

Kimi K2-Instruct-0905 is 3 months newer than Gemini 2.5 Flash-Lite.

Gemini 2.5 Flash-Lite

Jun 17, 2025

1.1 years ago

Kimi K2-Instruct-0905

Sep 5, 2025

10 months ago

2mo newer

Knowledge Cutoff

When training data ends

Gemini 2.5 Flash-Lite has a documented knowledge cutoff of 2025-01-01, while Kimi K2-Instruct-0905's cutoff date is not specified.

We can confirm Gemini 2.5 Flash-Lite's training data extends to 2025-01-01, but cannot make a direct comparison without Kimi K2-Instruct-0905's cutoff date.

Gemini 2.5 Flash-Lite

Jan 2025

Kimi K2-Instruct-0905

—

Outputs Comparison

Notice missing or incorrect data?Start an Issue discussion→

Key Takeaways

Google
Larger context window (1,048,576 tokens)
Supports multimodal inputs
Higher AIME 2025 score (49.8% vs 49.5%)
Higher Humanity's Last Exam score (5.1% vs 4.7%)
Moonshot AI
Higher Aider-Polyglot score (60.0% vs 26.7%)
Higher GPQA score (75.1% vs 64.6%)
Higher LiveCodeBench score (53.7% vs 33.7%)
Higher SimpleQA score (31.0% vs 10.7%)
Higher SWE-Bench Verified score (65.8% vs 31.6%)

Detailed Comparison

Interactive Arena

Judge for yourself.

Run your own prompts against Gemini 2.5 Flash-Lite and Kimi K2-Instruct-0905 side-by-side, then vote on the output you prefer.

Gemini 2.5 Flash-Lite
✓ Preferred
Kimi K2-Instruct-0905
Open in Playground
AI Model Comparison Table
Feature
Google
Gemini 2.5 Flash-Lite
Moonshot AI
Kimi K2-Instruct-0905

FAQ

Common questions about Gemini 2.5 Flash-Lite vs Kimi K2-Instruct-0905.

Which is better, Gemini 2.5 Flash-Lite or Kimi K2-Instruct-0905?

Kimi K2-Instruct-0905 shows notably better performance in the majority of benchmarks. Gemini 2.5 Flash-Lite is made by Google 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 Gemini 2.5 Flash-Lite compare to Kimi K2-Instruct-0905 in benchmarks?

Gemini 2.5 Flash-Lite scores FACTS Grounding: 84.1%, Global-MMLU-Lite: 81.1%, MMMU: 72.9%, GPQA: 64.6%, Vibe-Eval: 51.3%. 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 Gemini 2.5 Flash-Lite and Kimi K2-Instruct-0905?

Gemini 2.5 Flash-Lite supports 1.0M 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 Gemini 2.5 Flash-Lite and Kimi K2-Instruct-0905?

Key differences include multimodal support (yes vs no), licensing (Creative Commons Attribution 4.0 License vs MIT). See the full comparison above for benchmark-by-benchmark results.

Who makes Gemini 2.5 Flash-Lite and Kimi K2-Instruct-0905?

Gemini 2.5 Flash-Lite is developed by Google and Kimi K2-Instruct-0905 is developed by Moonshot AI.

Related comparisons