Benchmark of AI Models for 1C Development: Updated Ranking v2
Anthropic
OpenAI
Moonshot AI
xAI
Cursor
DeepSeek
Alibaba/Qwen
MiniMax
An updated benchmark of AI models for 1C development is presented. The testing involves 14 models, including Claude Opus 5, GPT 5.6 Sol, Kimi K3, and others. The leader is recognized as Claude Opus 5, while GPT 5.6 Sol performs best with MCP and instructions.
The author of a benchmark for 1С development has updated the model ranking to include 14 models: Claude Opus 5, Claude Fable 5, GPT 5.6 Sol, Kimi K3, Claude Opus 4.8, Grok 4.5, Composer 2.5, GPT 5.5, DeepSeek v4, GLM 5.2, Qwen 3.8 Max, Claude Sonnet 5, Mimi 2.5, and MiniMax M2.7. Testing was conducted in vendor environments (Claude Code, Codex, Kimi Code, Cursor, etc.) with maximum reasoning effort and a separate option using MCP servers for 1С. Tasks are divided into categories: algorithms, architecture, platform mechanisms, performance, metadata, and form development. Token consumption without caching is taken into account. According to the results, Claude Opus 5 is recognized as the best for coding but falls short of GPT 5.6 Sol in following instructions when using MCP. Kimi K3 showed a high level but slow speed. DeepSeek v4 and GLM 5.2 are recommended only for tight budgets and with MCP. Qwen 3.8 Max, Claude Sonnet 5, Mimi 2.5, and MiniMax M2.7 are not recommended for 1С. Plans include adding time measurement, total cost, and tasks for feature refinement. The author uses Cursor with Claude Opus for architecture, Grok/Composer for main tasks, and GPT 5.6 Sol for review and autonomous agents.
Source: Habr — хаб ИИ —
original