๐ง Models Directory
๐ก Use any model ID as the model parameter in your API calls. All models are accessible through /v1/chat/completions (text) unless noted otherwise.
DeepSeek deepseek.com
Frontier reasoning and general-purpose models from DeepSeek AI. Known for exceptional coding and reasoning capabilities.
| Model ID | Type | Description |
|---|---|---|
deepseek-chat | Text | General-purpose chat โ best all-around model for daily use |
deepseek-reasoner | Text | Extended reasoning / deep-thinking model โ excels at complex logic, math, and analysis |
deepseek-coder | Code | Code generation, debugging, and software engineering tasks |
Qwen (Tongyi Qianwen) qwen.alibaba.com
Alibaba Cloud's flagship model series. Qwen3-Max sets state-of-the-art benchmarks. Strong in multilingual tasks, especially Chinese-English.
| Model ID | Type | Description |
|---|---|---|
qwen-max | Text | Flagship โ Qwen3-Max, best quality, strongest reasoning |
qwen-plus | Text | Balanced performance and speed โ great value |
qwen-turbo | Text | Fast inference, cost-effective for high-volume use |
qwen-vl-plus | Vision | Multi-modal โ understands images, charts, and documents |
qwen-omni-flash | Text | Fast omnimodal model โ text, image, audio understanding |
Kimi (Moonshot AI) kimi.moonshot.cn
Moonshot AI's model family. Known for exceptional long-context handling (up to 2M tokens). K2 series features extended thinking.
| Model ID | Type | Description |
|---|---|---|
kimi-chat | Text | General conversation with ultra-long context support |
kimi-k2-thinking | Text | Extended thinking / reasoning โ designed for complex deep-dive analysis |
kimi-vl | Vision | Image understanding and visual reasoning |
ERNIE (Baidu) yiyan.baidu.com
Baidu's Wenxin Yiyan model family. ERNIE 4.0 is a top-tier model with strong Chinese-language performance and knowledge-grounded generation.
| Model ID | Type | Description |
|---|---|---|
ernie-4.0 | Text | Latest flagship โ Baidu's most capable model |
ernie-3.5 | Text | Previous generation โ solid, well-tested |
ernie-speed | Text | Fast inference, optimized for latency-sensitive apps |
ernie-lite | Text | Lightweight, most cost-effective ERNIE model |
GLM (Zhipu AI) zhipu.ai
Zhipu AI's GLM series. Strong multilingual capabilities and competitive performance. GLM-4V supports vision tasks.
| Model ID | Type | Description |
|---|---|---|
glm-4-plus | Text | Latest flagship โ Zhipu's most advanced model |
glm-4 | Text | Previous generation GLM-4 โ well-rounded general model |
glm-4-air | Text | Lightweight variant of GLM-4 โ fast and affordable |
glm-4v-plus | Vision | Multi-modal โ image understanding and analysis |
glm-4-flash | Text | Fast inference model โ optimized for speed |
Yi (01.AI) yi-01.ai
01.AI's open-source model family. Strong bilingual performance and active community ecosystem.
| Model ID | Type | Description |
|---|---|---|
yi-large | Text | Flagship โ 01.AI's most powerful general model |
yi-medium | Text | Balanced performance and speed |
yi-vision | Vision | Multi-modal โ understands and analyzes images |
MiniMax minimax.chat
Models from MiniMax, known for strong creative and conversational capabilities.
| Model ID | Type | Description |
|---|---|---|
minimax-text-01 | Text | General-purpose text generation and conversation |
minimax-vl-01 | Vision | Image understanding and multi-modal reasoning |
Baichuan AI baichuan-ai.com
Baichuan Intelligent Technology's model series. Strong performance in Chinese language tasks and vertical domains.
| Model ID | Type | Description |
|---|---|---|
baichuan4 | Text | Latest generation โ Baichuan's most capable model |
baichuan3-turbo | Text | Fast inference variant of Baichuan3 |
Spark (iFlytek) xfyun.cn
iFlytek's Spark (Xinghuo) model series. Strong in voice-related AI and educational domains. Developer platform: xfyun.cn.
| Model ID | Type | Description |
|---|---|---|
spark-4.0 | Text | Latest flagship โ iFlytek's most advanced model |
spark-3.5 | Text | Previous generation โ proven and stable |
spark-lite | Text | Lightweight โ fastest and most affordable Spark model |
Hunyuan (Tencent) hunyuan.tencent.com
Tencent's Hunyuan model series. Integrated with Tencent's ecosystem. Strong in Chinese social media context understanding.
| Model ID | Type | Description |
|---|---|---|
hunyuan-large | Text | Flagship โ Tencent's most capable general model |
hunyuan-standard | Text | Balanced performance for daily use |
hunyuan-vision | Vision | Multi-modal โ image understanding and reasoning |
Doubao (ByteDance) doubao.com
ByteDance's model series. Doubao powers many TikTok/ๆ้ณ AI features. Seed series represents ByteDance's frontier research.
| Model ID | Type | Description |
|---|---|---|
doubao-pro | Text | Latest flagship Doubao model |
doubao-lite | Text | Lightweight โ fast and cost-effective |
doubao-coding-plan | Code | Specialized for code generation and planning |
doubao-seed-1.8 | Text | Experimental frontier model โ ByteDance's latest research |
Model Selection Tips
General use: deepseek-chat, qwen-plus, glm-4 โ solid all-around performers
Complex reasoning: deepseek-reasoner, kimi-k2-thinking, qwen-max โ for math, logic, deep analysis
Coding: deepseek-coder, doubao-coding-plan โ code generation and engineering
Vision (image understanding): qwen-vl-plus, glm-4v-plus, kimi-vl, yi-vision โ send images + text prompts
High volume / budget: qwen-turbo, ernie-lite, spark-lite, doubao-lite โ fast and cheap
Get the Live Model List
The authoritative list of all currently available models can be fetched directly from the API:
curl https://www.tokencnn.com/v1/models \
-H "Authorization: Bearer sk-your-api-key"
New models are added frequently. Check this endpoint for the most up-to-date availability.