Models and Reasoning Effort

Overview

Qoder CLI CN offers three categories of model selection via /model, corresponding to the three tabs in the /model interface: Default, New Models, and Custom. For everyday use, we recommend the default Auto smart routing.

Ways to select a model

Default: Auto smart routing only

The Default tab in Qoder CLI CN offers only Auto (smart routing): it automatically selects the most suitable backend model for each task, balancing efficiency, output quality, and cost. It fits the vast majority of development scenarios and is the recommended everyday choice.

Difference from the Global edition: the Global edition's Default tab offers multiple tiered modes (such as Ultimate, Performance, etc.), while Qoder CLI CN provides only the single Auto mode. To use a specific model, choose one under "New Models" below.

New Models

In the New tab of /model you can pick a specific model. The currently available models are as follows (most support multimodal/vision):

Model	Description
Qwen3.8-Max	Qwen's latest base model with 2.4T parameters, leading in software engineering, office work, and deep reasoning
Qwen3.7-Max	Qwen's flagship model with top-tier agentic capabilities, able to complete long-horizon complex tasks autonomously
Qwen3.7-Plus	Comprehensively enhanced reasoning and agentic capabilities
Qwen3.6-Flash	Native vision-language model of the Qwen3.6 Flash series — fast and efficient
DeepSeek-V4-Pro	DeepSeek's latest model with outstanding agentic performance
DeepSeek-V4-Flash	DeepSeek's latest model — fast, low-cost, and well-balanced overall
GLM-5.2	Zhipu AI's flagship model, strong at complex systems engineering and long-horizon tasks
Kimi-K2.7-Code	Built for long-context coding: precise instruction following and reliable long-horizon task execution
MiniMax-M2.7	Native multimodal perception and frontier coding capabilities, suited for high-intensity workflows

Available models, their descriptions, and Credit consumption may change as the server-side model list is updated; always refer to the actual output of the /model interface and qodercn --list-models.

Custom models (Custom / BYOK)

If you need to use your own model subscription, you can connect a model provider in the Custom tab of /model and call third-party models with your own API Key. For full configuration instructions, see Custom Models.

Model parameter configuration

Some models support configurable parameters to better fit different tasks. Available parameters depend on the current model; the CLI shows the options supported by the current model in the /model list and the model parameter panel.

Context window

The context window controls the maximum context length available to the conversation. Common options include:

Option	Description
200K	Standard context window, sufficient for most tasks
400K	Extended context, for larger codebases or longer conversations
1M	Maximum context, for very large projects

In TUI mode, type /context-window to open the model parameter panel and adjust the context window.

Reasoning effort

Controls how deeply the model reasons before generating a response. Available options vary by model; after selecting a model, the interface shows the levels that model supports.

Option	Description
low	Minimal reasoning, fastest response
medium	Moderate reasoning depth
high	Deep reasoning, for complex tasks
xhigh	In-depth analysis, for difficult problems
max	Maximum reasoning depth, for the most complex challenges

In TUI mode, type /effort to adjust the current model's reasoning effort. Without arguments it opens the model parameter panel; you can also specify a level directly:

/effort high
/effort auto --session-only
/effort off

By default the setting is saved to ~/.qoder-cn/settings.json; add --session-only to apply it to the current session only.

Fast mode

Some models offer a Fast mode that prioritizes response speed. Credit consumption may change when enabled; refer to the model parameter panel for details.

/fast on

For which parameters a given model supports, refer to what the /model parameter panel actually shows.

Switching models

Switch in the TUI

In TUI mode, type /model to open the selection interface, which is divided into three tabs — Default, New, and Custom:

/model

Use ↑ / ↓ to select, Tab to switch tabs, Enter to confirm, and Esc to close. The selected model is automatically saved to ~/.qoder-cn/settings.json, so you don't need to choose again next time you launch.

Specify via command-line flags

Use the --model, --reasoning-effort, and --context-window flags to specify directly at startup, effective for this session only. Default and New Models use the model name; Custom uses the modelID:

qodercn --model auto            # Use Auto smart routing
qodercn --model Qwen3.8-Max     # Specify a model from the New tab
qodercn --reasoning-effort high
qodercn --context-window 400000

List available models

Use --list-models to print the models available to the current account in headless mode — handy for scripts and automation:

qodercn --list-models

Adding a custom model

Type /model and switch to the Custom tab
Select Add custom model... and press Enter
Follow the guide to choose Provider → model type → specific model
Fill in the API Key and other configuration
After validation passes, it is saved automatically and can be switched to immediately

To delete a custom model: select the target model in the Custom tab and press d.

​Overview

​Ways to select a model

​Default: Auto smart routing only

​New Models

​Custom models (Custom / BYOK)

​Model parameter configuration

​Context window

​Reasoning effort

​Fast mode

​Switching models

​Switch in the TUI

​Specify via command-line flags

​List available models

​Adding a custom model