whichLlmmodel
Back to Dashboard

Moonshot AI (Kimi)kimi-k2.5VSOpenAIGPT-5 mini

Analysis by:the whichllmmodel Editorial Team|Updated: June 2026

Our Take

We recommend kimi-k2.5 for its clear benchmark advantage, or the 1.7x cheaper GPT-5 mini only if your budget requires optimizing costs for very high-volume pipelines. While kimi-k2.5 offers superior reasoning and coding, it carries a moderate price premium. Choose kimi-k2.5 for quality, or GPT-5 mini for cost optimization.
WHY?
Benchmark Calculations & Evidence:
  • Coding Benchmarks: Both models were evaluated on the SWE-bench Pro benchmark. kimi-k2.5 scored 50.7%, while GPT-5 mini scored 45.7%.
  • Reasoning Benchmarks: Both models were evaluated on the GPQA Diamond benchmark. kimi-k2.5 scored 87.6%, while GPT-5 mini scored 81.6%.
  • Cost Efficiency: GPT-5 mini pricing ($0.25/M input, $2/M output) is 1.7x cheaper than kimi-k2.5 ($0.6/M input, $3/M output).
  • Was this recommendation helpful?
    Model Specs

    kimi-k2.5

    Open SourceAPI Available

    Benchmarks & Scores

    Coding (swe-bench-pro)Winner (+5.0%)
    50.7%

    complex codebases, multi-file repositories, and architectural planning

    Reasoning (gpqa-diamond)Winner (+6.0%)
    87.6%

    graduate-level science QA

    Cost & Context

    Cost (per 1M tokens)
    $1.20Input: $0.60 | Output: $3.00
    Context Window
    262.14k tokens
    Model Specs

    GPT-5 mini

    Benchmarks & Scores

    Coding (swe-bench-pro)
    45.7%

    complex codebases, multi-file repositories, and architectural planning

    Reasoning (gpqa-diamond)
    81.6%

    graduate-level science QA

    Cost & Context

    Cost (per 1M tokens)1.7x cheaper
    $0.69Input: $0.25 | Output: $2.00
    Context WindowLarger
    400k tokens

    Frequently Asked Questions about kimi-k2.5 vs GPT-5 mini

    GPT-5 mini is cheaper than kimi-k2.5. GPT-5 mini has a blended cost of $0.69/1M tokens, which is about 1.7x cheaper than kimi-k2.5 at $1.20/1M tokens.

    kimi-k2.5 is better for coding tasks on this benchmark. It scores 50.7% on swe-bench-pro (complex codebases, multi-file repositories, and architectural planning) compared to GPT-5 mini which scores 45.7%.

    Do you want to find a model for your constraints?

    Use our interactive model finder to filter LLMs by reasoning capability, coding performance, cost, and context length.

    Open Model Finder