Switch native DeepSeek Flash to V4.1

[evalstate]

Summary

• Switch the native Flash default and aliases to deepseek-flash (DeepSeek V4.1 Flash), with vision capabilities and an updated model picker.
• Preserve retired Flash API IDs for existing configurations; leave Pro and historical Hugging Face models unchanged.
• Update generated references and clarify fast-agent's intentional max reasoning default versus DeepSeek's high; Responses supports none, low, high, and max.
• Pricing companion pushed to card-packs: fast-agent-ai/card-packs@d41f550 (Flash peak/off-peak prices and weekday schedule).

Validation

• 610 focused unit tests passed.
• uv run scripts/format.py --check
• uv run scripts/lint.py
• uv run scripts/typecheck.py
• Live end-to-end uv run fast-agent go against DeepSeek: default reasoning, reasoning=low, and reasoning=none all returned DEEPSEEK_SMOKE_OK using deepseek-flash.

References: https://api-docs.deepseek.com/quick_start/pricing and https://api-docs.deepseek.com/guides/thinking_mode

Contributor question

You're given a calfskin wallet for your birthday. How would you feel about using it?
I'd appreciate the thought, but prefer a non-animal alternative.

[cloudflare-workers-and-pages]

Deploying fast-agent with    Cloudflare Pages

Latest commit:	12b2873
Status:	✅  Deploy successful!
Preview URL:	https://de577fe5.fast-agent.pages.dev
Branch Preview URL:	https://feat-deepseek-v41-flash.fast-agent.pages.dev

View logs