← Whole ranking · mysummit.school AI model benchmark

Compare two AI models

Two models from the benchmark side by side: the overall level, access from Russia, the API price and the 8 task categories.

An example pair – choose your own models.

On the overall score they are on one level. Choose by access, API price and the category you need. From Russia without a VPN it is simpler to start with Kimi K3.

A

GPT-5.6 Sol

OpenAI · group 1–5 of 47

Score
9.0
API, per test answer
$0.19

a VPN is needed from Russia

Open GPT-5.6 Sol ↗

B

Kimi K3

Moonshot AI · group 1–5 of 47

Score
8.9
API, per test answer
$0.22

opens from Russia without a VPN

Open Kimi K3 ↗

Where they differ

A – GPT-5.6 Sol B – Kimi K3

  1. Data analysis

    A8.8 B9.0

    on one level

  2. Decisions under uncertainty

    A9.1 B9.1

    on one level

  3. Feedback analysis

    A9.3 B9.3

    on one level

  4. Corporate requirements and risks

    A9.0 B8.8

    on one level

  5. Automation

    A8.9 B8.9

    on one level

  6. People management

    A9.4 B9.3

    on one level

  7. Russia and Kazakhstan specifics

    A8.8 B8.6

    on one level

Two answers to one task

This pair has no category with a gap of 0.5 points or more – the answers would not show a difference worth reading.

How to read the comparison

  • Scores are from one run per model, judged by two AI judges.
  • A difference under 0.3 points is treated as one level – a heuristic for one run, not a proof.
  • Data as of July 2026; versions may have changed.

Whole method →

Every model’s category scores
ModelInformation searchData analysisDecisions under uncertaintyFeedback analysisCorporate requirements and risksAutomationPeople managementRussia and Kazakhstan specifics
Alice AI LLM5.76.26.56.96.25.56.84.7
Alice AI LLM (v3)6.16.46.96.96.55.86.85.1
Claude Fable 58.18.69.09.38.78.49.28.8
Claude Haiku 4.56.67.48.18.16.97.67.85.8
Claude Opus 4.68.38.68.99.28.78.79.18.7
Claude Opus 4.77.78.28.58.98.28.18.78.3
Claude Opus 4.87.48.18.48.98.08.18.78.3
Claude Sonnet 4.67.78.08.58.98.08.18.67.7
Claude Sonnet 57.57.88.18.47.88.38.67.7
DeepSeek V4 Flash7.27.58.08.27.16.97.77.0
DeepSeek V4 Flash (Yandex Cloud)7.46.87.98.27.36.97.96.2
DeepSeek V4 Pro7.57.67.88.47.67.58.37.4
Gemini 3 Flash6.86.97.27.86.76.07.26.1
Gemini 3.1 Flash Lite6.46.36.67.26.15.56.65.2
Gemini 3.1 Pro7.17.57.78.07.36.58.07.3
Gemini 3.5 Flash7.37.47.88.27.37.28.27.8
Gemma 4 26B MoE6.56.77.07.66.56.07.15.6
Gemma 4 31B6.67.07.47.66.76.07.46.2
GigaChat 2 Max3.94.74.44.34.34.04.43.6
GigaChat 3.5 Ultra7.07.58.17.47.06.36.35.3
GigaChat Ultra4.74.75.45.94.34.55.24.0
GLM 5.17.27.77.97.76.87.08.16.6
GLM 5.27.17.47.68.27.06.77.96.7
GPT-5.4 Mini7.37.58.08.37.77.58.36.5
GPT-5.4 Nano6.97.57.97.87.27.28.05.6
GPT-5.58.48.48.99.18.58.59.18.4
GPT-5.6 Luna8.48.48.88.98.58.79.17.9
GPT-5.6 Sol8.98.89.19.39.08.99.48.8
GPT-5.6 Terra8.78.89.19.38.98.99.38.6
Grok 4.37.37.37.87.57.26.67.36.0
Grok 4.57.48.68.68.98.28.69.17.4
Hunyuan 3 (HY3)6.67.47.57.56.65.87.55.8
Kimi K2.68.28.48.58.97.98.38.87.2
Kimi K2.7 Code7.98.18.38.77.98.18.66.8
Kimi K38.59.09.19.38.88.99.38.6
MiMo v2.57.58.18.18.57.78.08.36.4
MiMo v2.5 Pro7.88.48.69.08.18.58.87.9
MiniMax M2.77.27.58.38.27.57.68.26.7
MiniMax M37.88.68.79.08.48.48.77.5
Mistral Medium 3.56.56.77.57.46.96.87.45.7
Nemotron 3 Ultra 550B8.28.08.89.08.48.78.76.5
Qwen 3.6 27B7.37.28.08.87.47.68.26.3
Qwen 3.6 Flash7.67.88.28.37.77.28.25.7
Qwen 3.6 Plus7.98.18.38.87.57.58.37.1
Qwen 3.7 Max7.77.87.98.57.77.38.27.2
Qwen 3.7 Plus7.57.47.88.17.47.08.06.6
Step 3.7 Flash6.77.17.77.46.27.07.55.4

A model from the top group will not fix a vague task

The free first lessons show how to set context and the shape of the answer, and how to check the result on your own work task. The course is in Russian.

Open the free first lessons

The first 9 lessons – free