[Bug report] Update DeepSeek models to V4

[starloreh]

Basic information

• Operating System: Bluefin Linux
• Calibre Version: 9.7
• Plugin Version: ebook-translator_20260101235654
• Plugin Installation: From Rolling Release

Describe the bug
deepseek (chat) only offers the old models deepseek-chat and deepseek-reasoner
Expected behavior
Offer the deepseek v4 models: DeepSeek-V4-Flash, DeepSeek-V4-Pro
DeepSeek API docs say:
The model names deepseek-chat and deepseek-reasoner will be deprecated in the future. For compatibility, they correspond to the non-thinking mode and thinking mode of deepseek-v4-flash, respectively.

So the real problem is the current non availability of DeepSeek-V4-Pro among the models offered on the plugin.

[VolatilCapital]

For reference, the current DeepSeek model identifiers per the official docs:

• deepseek-v4-flash — fast model, 1M context window
• deepseek-v4-pro — advanced model, 1M context window (75% discount through 2026-05-31)

The legacy aliases deepseek-chat (non-thinking) and deepseek-reasoner (thinking) currently map to deepseek-v4-flash and are scheduled for deprecation on 2026-07-24.

A minimal fix in engines/deepseek.py would be to expose the new identifiers in the models list:

models: list[str] = [
    'deepseek-v4-flash',
    'deepseek-v4-pro',
    'deepseek-chat',      # deprecated 2026-07-24
    'deepseek-reasoner',  # deprecated 2026-07-24
]

Docs: https://api-docs.deepseek.com/quick_start/pricing

As a workaround in the meantime, the built-in Custom Engine works out of the box with DeepSeek V4 since the API is OpenAI-compatible (endpoint https://api.deepseek.com/v1/chat/completions, model: deepseek-v4-flash, response parser response['choices'][0]['message']['content']).