CowAgent - CowAgent

Besides chat, the same key covers image understanding, image generation, speech, embeddings and web search. Add it once in the Web Console and every capability below is ready to use. CowAgent runs two sites with separate accounts. A key only works on the site that issued it:

Site	API Base	Console (create API keys)
Global	`https://api.cowagent.ai/v1`	cloud.cowagent.ai
China	`https://api.cowagent.cn/v1`	cloud.cowagent.cn

The easiest way is the “Model Management” page of the Web Console: the CowAgent provider dialog comes with the Global API Base filled in, so you only paste your key. Every capability below reuses the same key.

Text Chat

{
  "bot_type": "cowagent",
  "model": "deepseek-v4.1-flash",
  "cowagent_api_key": "YOUR_API_KEY",
  "cowagent_api_base": "https://api.cowagent.ai/v1"
}

Parameter	Description
`bot_type`	Set to `cowagent`
`cowagent_api_key`	Create one in the Global console
`cowagent_api_base`	`https://api.cowagent.ai/v1` for a Global key, `https://api.cowagent.cn/v1` for a China key. When omitted it follows the UI language: the China site for Chinese, the Global site otherwise
`model`	One of the model IDs below

Available models: gpt-6.1-terra, gpt-6.1-sol, gpt-6-luna, gpt-6-astra, claude-opus-5-5, claude-sonnet-5-5, claude-haiku-5-5, claude-fable-5-1, gemini-4-argon, gemini-3.8-flash, deepseek-v4.1-flash, deepseek-v4-pro, deepseek-v4-flash, qwen3.8-flash, qwen3.8-max, qwen3.6-plus, glm-5.3-flash, glm-5.3, kimi-k3.1-pro, kimi-k3, MiniMax-M3.1-Flash, MiniMax-M3, doubao-seed-2-1-turbo. See the console for the full list and pricing.

Image Understanding

No extra setup once the key is configured: images go to the main model when it accepts image input, otherwise the Vision tool uses deepseek-v4.1-flash. To pin a model:

{
  "tools": {
    "vision": {
      "provider": "cowagent",
      "model": "deepseek-v4.1-flash"
    }
  }
}

Models with image input: gpt-6.1-terra, gpt-6.1-sol, gpt-6-luna, gpt-6-astra, claude-fable-5-1, gemini-4-argon, deepseek-v4.1-flash, qwen3.8-flash, qwen3.8-max, qwen3.6-plus, glm-5.3-flash, MiniMax-M3.1-Flash, doubao-seed-2-1-turbo.

Image Generation

{
  "skills": {
    "image-generation": {
      "provider": "cowagent",
      "model": "gpt-image-2.5-flare"
    }
  }
}

Model ID	Notes
`gpt-image-2.5-flare`	OpenAI, default
`gpt-image-2.5-sunburst`	OpenAI, higher editing precision
`gemini-3.1-flash-image`	Nano Banana 2
`gemini-3-pro-image`	Nano Banana Pro
`seedream-5.0-pro`	ByteDance Doubao Seedream
`seedream-5.0-lite`	ByteDance Doubao Seedream

Speech-to-Text (ASR)

{
  "voice_to_text": "cowagent",
  "voice_to_text_model": "seed-asr"
}

Text-to-Speech (TTS)

{
  "text_to_voice": "cowagent",
  "text_to_voice_model": "seed-tts-2.0",
  "tts_voice_id": "en_female_dacey_uranus_bigtts"
}

`tts_voice_id`	Voice
`zh_female_vv_uranus_bigtts`	Chinese · Female · Vivi (default)
`zh_female_xiaohe_uranus_bigtts`	Chinese · Female · Xiaohe
`zh_male_m191_uranus_bigtts`	Chinese · Male · Yunzhou
`zh_male_taocheng_uranus_bigtts`	Chinese · Male · Xiaotian
`zh_male_liufei_uranus_bigtts`	Chinese · Male · Liufei
`en_male_tim_uranus_bigtts`	English · Male · Tim
`en_female_dacey_uranus_bigtts`	English · Female · Dacey
`en_female_stokie_uranus_bigtts`	English · Female · Stokie

Embedding

{
  "embedding_provider": "cowagent",
  "embedding_model": "text-embedding-3-small"
}

The default model is text-embedding-3-small (1536 dimensions). After changing the embedding, run /memory rebuild-index to rebuild the index.

Web Search

Once the key is configured, the web_search tool can search through CowAgent with no extra setup. To always use it:

{
  "tools": {
    "web_search": {
      "strategy": "fixed",
      "provider": "cowagent"
    }
  }
}

​Text Chat

​Image Understanding

​Image Generation

​Speech-to-Text (ASR)

​Text-to-Speech (TTS)

​Embedding

​Web Search

Text Chat

Image Understanding

Image Generation

Speech-to-Text (ASR)

Text-to-Speech (TTS)

Embedding

Web Search