Qoder CN CLI comes with built-in domestic AI models and provides a flexible selection mechanism to help you find the best balance between development efficiency, output quality, and cost.
The tiered model selector provides developers with five high-performance model pools that strike different balances between cost and performance. Like the driving modes of a smart car, it selects the most appropriate gear for each task.
In TUI mode, type
Controls how deeply the model reasons before generating a reply. The available options vary by model; after you select a model, the interface shows the tiers it supports.
In TUI mode, type
By default, the setting is saved to
Some models offer a Fast mode that prioritizes response speed. When enabled, Credit consumption may change; refer to the model parameter panel for details.
Currently Ultimate, Performance, DeepSeek-V4-Pro, DeepSeek-V4-Flash, Qwen3.7-Max, Qwen3.7-Plus, GLM-5.2, MiniMax-M3, and Kimi-K2.7-Code support parameter configuration. Among them, Qwen3.7-Max, Qwen3.7-Plus, GLM-5.2, and MiniMax-M3 only support the context window; Kimi-K2.7-Code only supports enabling Fast mode; the remaining models support both context window and reasoning effort.
In TUI mode, type
Use
Use the
Use
Model Selection
Tiered Models (Default)
The tiered model selector provides developers with five high-performance model pools that strike different balances between cost and performance. Like the driving modes of a smart car, it selects the most appropriate gear for each task.
Applies to: Tier; Description; Use Cases; Credit Consumption; Auto (Smart Routing); Intelligently selects the most suitable model, balancing performance and cost; Most everyday development work, recommended as the default; ~1.0×; Ultimate; Expert-level deep reasoning and thinking capability; Complex system design, difficult problem analysis; ~1.6×; Performance; Advanced reasoning capability, high-quality output; Core feature implementation, architecture design, code refactoring; ~1.1×; Efficient; Standard reasoning capability, high cost-effectiveness; Basic code generation, unit tests, everyday Q&A; ~0.3×; Lite; Basic reasoning capability, free to use; Quick validation, basic logic implementation, quick Q&A; Free; Model tier; Credit consumption multiplier; Consumption per example task; Auto; ~1.0×; 10 Credits; Ultimate; ~1.6×; 20 Credits; Performance; ~1.1×; 11 Credits; Efficient; ~0.3×; 3 Credits; Model name; Model overview; Credit consumption multiplier; Qwen3.7-Max; A comprehensive leap in Agentic capability, able to autonomously complete long, complex tasks; 0.5×; Qwen3.7-Plus; A comprehensive leap in reasoning capability, efficiency, and multimodal experience; 0.1×; DeepSeek-V4-Pro; Excels at complex reasoning, code generation, and engineering tasks; 0.5×; DeepSeek-V4-Flash; Fast reasoning and low cost, with balanced overall capabilities; 0.1×; GLM-5.2; Excels at complex system engineering and long-running tasks; 0.6×; Kimi-K2.7-Code; Kimi's latest model, purpose-built for long-context coding: precise instruction following and reliable long-running task execution; 0.3×; MiniMax-M3; Native multimodal perception, cutting-edge coding capability, and 1M context depth, suitable for high-intensity workflows; 0.2×; Option; Description; 200K; Standard context window, covers most task needs; 400K; Extended context, suitable for larger codebases or longer conversations; 1M; Maximum context, suitable for ultra-large-scale projects
/context-window to open the model parameter panel and adjust the context window.
Reasoning Effort
Controls how deeply the model reasons before generating a reply. The available options vary by model; after you select a model, the interface shows the tiers it supports.
| Option | Description |
|---|---|
| low | Minimal reasoning, fastest response |
| medium | Moderate reasoning depth |
| high | Deep reasoning, suitable for complex tasks |
| xhigh | In-depth analysis, suitable for highly difficult problems |
| max | Maximum reasoning depth, suitable for the most complex challenges |
/effort to adjust the reasoning effort of the current model. Without arguments, it opens the model parameter panel; you can also specify the level directly:
~/.qoder-cn/settings.json; adding --session-only applies it only to the current session.
Fast Mode
Some models offer a Fast mode that prioritizes response speed. When enabled, Credit consumption may change; refer to the model parameter panel for details.
Supported Models
Currently Ultimate, Performance, DeepSeek-V4-Pro, DeepSeek-V4-Flash, Qwen3.7-Max, Qwen3.7-Plus, GLM-5.2, MiniMax-M3, and Kimi-K2.7-Code support parameter configuration. Among them, Qwen3.7-Max, Qwen3.7-Plus, GLM-5.2, and MiniMax-M3 only support the context window; Kimi-K2.7-Code only supports enabling Fast mode; the remaining models support both context window and reasoning effort.
Switching Models
Switch in TUI
In TUI mode, type /model to open the selection interface, which is divided into three tabs: Default, New Models, and Custom:
↑ / ↓ to navigate, Tab to switch tabs, Enter to confirm, and Esc to close.
The selected model is automatically saved to ~/.qoder-cn/settings.json and does not need to be selected again on the next launch.
Command-Line Arguments
Use the --model, --reasoning-effort, and --context-window arguments to specify model settings directly at startup. These only apply to the current session:
List Available Models
Use --list-models to output the list of models available to the current account in a non-interactive mode. This is suitable for scripts or automation scenarios:
Adding a Custom Model
- Type
/modeland switch to the Custom tab - Select
Add custom model...and pressEnter - Follow the prompts to select Provider → Model Type → Model
- Fill in the API Key and other configuration information
- After verification, it is automatically saved and can be used immediately
d.