Skip to content

feat(workflow): upgrade default MiniMax model to M3 #66

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
octo-patch wants to merge 1 commit into
base: main
Choose a base branch
from
Open

feat(workflow): upgrade default MiniMax model to M3 #66

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions workflow/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -43,7 +43,7 @@ Edit `config.py` to set:
- **API Keys** (optional, for Full Workflow):
- `GEMINI_API_KEY` + `GEMINI_API_URL`: Gemini API (recommended; fill base URL when using proxy)
- `DASHSCOPE_API_KEY` + `DASHSCOPE_BASE_URL`: Qwen/DashScope API (base URL for region, e.g. Singapore/US)
- `MINIMAX_API_KEY`: [MiniMax](https://platform.minimaxi.com/) API for prompt generation (OpenAI-compatible; uses MiniMax-M2.7 by default)
- `MINIMAX_API_KEY`: [MiniMax](https://platform.minimaxi.com/) API for prompt generation (OpenAI-compatible; uses MiniMax-M3 by default)

> **Note (API Key Fallback):** If you do **not** provide a Gemini API key, the workflow will automatically use the Qwen (DashScope) API key to call qwen-plus and Z-Image for prompt generation and first frame generation. Alternatively, you can use a **MiniMax API key** (`MINIMAX_API_KEY`) for prompt rewriting and image prompt generation via MiniMax's OpenAI-compatible API. **We still recommend using a Gemini API key** for best quality.

Expand Down Expand Up @@ -115,7 +115,7 @@ export MOVA_MODEL_PATH="/path/to/your/MOVA-360p-hf" # 或 MOVA-720p-hf
- **API 密钥**（可选，完整工作流需要）：
- `GEMINI_API_KEY` + `GEMINI_API_URL`：Gemini API（推荐；使用代理时需填写 base URL）
- `DASHSCOPE_API_KEY` + `DASHSCOPE_BASE_URL`：通义千问 / DashScope API（base URL 可选，用于指定地域如新加坡/美国）
- `MINIMAX_API_KEY`：[MiniMax](https://platform.minimaxi.com/) API，用于提示词生成（OpenAI 兼容接口；默认使用 MiniMax-M2.7）
- `MINIMAX_API_KEY`：[MiniMax](https://platform.minimaxi.com/) API，用于提示词生成（OpenAI 兼容接口；默认使用 MiniMax-M3）

> **说明（API 密钥回退）：** 若不提供 Gemini API 密钥，工作流会自动使用 Qwen（DashScope）API 密钥调用 qwen-plus 和 Z-Image 完成提示词生成和首帧图生成。也可使用 **MiniMax API 密钥**（`MINIMAX_API_KEY`）通过 MiniMax 的 OpenAI 兼容 API 完成提示词改写和首帧图提示词生成。**我们仍推荐使用 Gemini API 密钥**以获得更好效果。

Expand Down
2 changes: 1 addition & 1 deletion workflow/config.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -89,7 +89,7 @@

MINIMAX_MODEL = os.environ.get(
'MINIMAX_MODEL',
'MiniMax-M2.7'
'MiniMax-M3'
)

# ============================================================================
Expand Down
4 changes: 2 additions & 2 deletions workflow/generate_first_frame.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -201,7 +201,7 @@ def generate_image_prompt_qwen(
def generate_image_prompt_minimax(
user_input: str,
api_key: str,
model: str = "MiniMax-M2.7",
model: str = "MiniMax-M3",
base_url: str = None,
) -> str:
"""
Expand All @@ -210,7 +210,7 @@ def generate_image_prompt_minimax(
Args:
user_input: 用户原始输入
api_key: MiniMax API Key
model: 模型名称，默认 MiniMax-M2.7
model: 模型名称，默认 MiniMax-M3
base_url: API base URL（可选，默认 https://api.minimax.io/v1）

Returns:
Expand Down
6 changes: 3 additions & 3 deletions workflow/prompt_rewriter_with_image.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -266,7 +266,7 @@ def generate_video_description_minimax(
user_input: str,
first_frame_elements: str,
api_key: str,
model: str = "MiniMax-M2.7",
model: str = "MiniMax-M3",
base_url: str = None,
) -> str:
"""
Expand All @@ -277,7 +277,7 @@ def generate_video_description_minimax(
user_input: 用户原始输入
first_frame_elements: 首帧图视觉元素描述
api_key: MiniMax API Key
model: 模型名称，默认 MiniMax-M2.7
model: 模型名称，默认 MiniMax-M3
base_url: API base URL（可选，默认 https://api.minimax.io/v1）

Returns:
Expand Down Expand Up @@ -365,7 +365,7 @@ def main():
'--model',
type=str,
default=None,
help='模型名称（Gemini 默认: gemini-2.5-pro；Qwen 默认: qwen-plus；MiniMax 默认: MiniMax-M2.7）'
help='模型名称（Gemini 默认: gemini-2.5-pro；Qwen 默认: qwen-plus；MiniMax 默认: MiniMax-M3）'
)
parser.add_argument(
'--minimax-api-key',
Expand Down
6 changes: 3 additions & 3 deletions workflow/tests/test_minimax_provider.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -82,21 +82,21 @@ def test_minimax_config_defaults(self):
from config import MINIMAX_API_KEY, MINIMAX_BASE_URL, MINIMAX_MODEL
self.assertIsInstance(MINIMAX_API_KEY, str)
self.assertEqual(MINIMAX_BASE_URL, 'https://api.minimax.io/v1')
self.assertEqual(MINIMAX_MODEL, 'MiniMax-M2.7')
self.assertEqual(MINIMAX_MODEL, 'MiniMax-M3')

def test_minimax_config_from_env(self):
with patch.dict(os.environ, {
'MINIMAX_API_KEY': 'test_key_123',
'MINIMAX_BASE_URL': 'https://custom.api/v1',
'MINIMAX_MODEL': 'MiniMax-M2.5',
'MINIMAX_MODEL': 'MiniMax-M2.7',
}):
# Re-import to pick up env vars
import importlib
import config
importlib.reload(config)
self.assertEqual(config.MINIMAX_API_KEY, 'test_key_123')
self.assertEqual(config.MINIMAX_BASE_URL, 'https://custom.api/v1')
self.assertEqual(config.MINIMAX_MODEL, 'MiniMax-M2.5')
self.assertEqual(config.MINIMAX_MODEL, 'MiniMax-M2.7')
# Reload with clean env
for key in ['MINIMAX_API_KEY', 'MINIMAX_BASE_URL', 'MINIMAX_MODEL']:
os.environ.pop(key, None)
Expand Down