Supported Models

KorProxy supports all CLIProxyAPI models. Click the copy button to copy model IDs for use in Cline, Cursor, Windsurf, or any OpenAI-compatible tool.

Showing 39 of 39 models

How to Use Model IDs

Copy any model ID and use it in your AI coding tool's configuration. For example, in Cline or Cursor settings:

# Example configurations

model: claude-opus-4-5-20251101

model: gpt-5.1-codex-max

model: gemini-2.5-pro

baseUrl: http://localhost:1337/v1

Thinking/Reasoning: For OpenAI Codex models, append reasoning level in parentheses: gpt-5.1-codex-max(high). Claude models have built-in thinking budgets. Gemini models use thinkingBudget parameter.

Gemini 3 Pro Preview

gemini-3-pro-preview

Next-generation flagship model with advanced reasoning

1M+ contextThinking supportMultimodal

Gemini 3 Pro Image Preview

gemini-3-pro-image-preview

Gemini 3 Pro with image generation capabilities

1M+ contextImage generationMultimodal

Gemini 2.5 Pro

gemini-2.5-pro

Stable flagship model for complex tasks

1M contextThinking supportBest for complex tasks

Gemini 2.5 Flash

gemini-2.5-flash

Fast and efficient mid-size model

1M contextFast responsesCost-effective

Gemini 2.5 Flash Lite

gemini-2.5-flash-lite

Smallest and most cost effective model

1M contextUltra-fastBest value

Gemini 2.5 Flash Image

gemini-2.5-flash-image

State-of-the-art image generation and editing

Image generationImage editingMultimodal

Gemini Pro Latest

gemini-pro-latest

Always points to latest Pro release

Auto-updated1M contextThinking support

Gemini Flash Latest

gemini-flash-latest

Always points to latest Flash release

Auto-updatedFastCost-effective

Claude 4.5 Opus

claude-opus-4-5-20251101

Premium model combining maximum intelligence with practical performance

200K context64K outputThinking: 1024-100000

Claude 4.5 Sonnet

claude-sonnet-4-5-20250929

Balanced model with excellent coding capabilities

200K context64K outputThinking: 1024-100000

Claude 4.5 Haiku

claude-haiku-4-5-20251001

Fastest Claude model for quick tasks

200K context64K outputNo thinking

Claude 4.1 Opus

claude-opus-4-1-20250805

Previous generation Opus model

200K context32K outputThinking support

Claude 4 Opus

claude-opus-4-20250514

Claude 4 flagship model

200K context32K outputThinking support

Claude 4 Sonnet

claude-sonnet-4-20250514

Claude 4 balanced model

200K context64K outputThinking support

Claude 3.7 Sonnet

claude-3-7-sonnet-20250219

Claude 3.7 generation Sonnet

128K context8K outputThinking support

Claude 3.5 Haiku

claude-3-5-haiku-20241022

Claude 3.5 generation Haiku

128K context8K outputNo thinking

GPT 5.1 Codex Max

gpt-5.1-codex-max

Most powerful Codex model for complex agentic tasks

400K context128K outputLevels: low, medium, high, xhigh

GPT 5.1 Codex

gpt-5.1-codex

Latest Codex model for coding tasks

400K context128K outputLevels: low, medium, high

GPT 5.1 Codex Mini

gpt-5.1-codex-mini

Smaller, faster version of Codex

400K contextCost-effectiveLevels: low, medium, high

GPT 5.1

gpt-5.1

Latest GPT model for general tasks

400K context128K outputMultimodal

GPT 5.2

gpt-5.2

Next generation GPT model

400K context128K outputLatest

GPT 5 Codex

gpt-5-codex

GPT 5 optimized for code

400K contextCode-focusedLevels: low, medium, high

GPT 5

gpt-5

GPT 5 base model

400K context128K outputLevels: minimal, low, medium, high

Qwen3 Coder Plus

qwen3-coder-plus

Advanced code generation and understanding model

32K context8K outputMulti-language

Qwen3 Coder Flash

qwen3-coder-flash

Fast code generation model

8K contextFastCost-effective

Qwen3 Vision Model

vision-model

Multimodal vision model

32K contextVisionMultimodal

TStars 2.0

tstars2.0

iFlow TStars multimodal assistant

MultimodalAssistantGeneral

Qwen3 Coder Plus

qwen3-coder-plus

Qwen3 Coder Plus via iFlow

Code generationMulti-languageFast

Qwen3 Coder 480B

qwen3-coder

Qwen3 Coder 480B A35B

Large modelCode-focusedPowerful

Qwen3 Max

qwen3-max

Qwen3 flagship model

FlagshipGeneral purposePowerful

Qwen3 VL Plus

qwen3-vl-plus

Qwen3 multimodal vision-language

VisionMultimodalAnalysis

Kimi K2

kimi-k2

Moonshot Kimi K2 general model

General purposeFastEfficient

Kimi K2 Thinking

kimi-k2-thinking

Moonshot Kimi K2 with extended thinking

Extended thinkingReasoningAnalysis

GLM 4.6

glm-4.6

Zhipu GLM 4.6 general model

General purposeChineseFast

DeepSeek V3.2

deepseek-v3.2

DeepSeek V3.2 experimental

ExperimentalPowerfulReasoning

DeepSeek V3.1 Terminus

deepseek-v3.1

DeepSeek V3.1 Terminus

StableGeneral purposeFast

DeepSeek R1

deepseek-r1

DeepSeek reasoning model R1

ReasoningAnalysisProblem-solving

DeepSeek V3 671B

deepseek-v3

DeepSeek V3 large model

671B parametersPowerfulGeneral

MiniMax M2

minimax-m2

MiniMax M2 model

General purposeFastEfficient

API Endpoints

KorProxy runs on http://localhost:1337 and provides these endpoints:

OpenAI/v1/chat/completions

OpenAI/v1/responses

Claude/v1/messages

Gemini/v1beta/models/:generateContent

Models/v1/models

Multi-Account Load Balancing

KorProxy supports multiple accounts per provider with round-robin load balancing. Add multiple accounts to distribute requests and avoid rate limits.