Santosh/multimodal

[santoshkumarradha]

Summary

This PR adds comprehensive multimodal support to the AgentField SDK with a unified UX pattern. All media generation (image, audio, video, transcription) works through consistent .ai_*() methods with automatic provider routing based on model prefix.

Key Features

Unified Multimodal UX

# Image generation - automatic provider routing
result = await app.ai_generate_image("A sunset", model="fal-ai/flux/dev")      # → Fal
result = await app.ai_generate_image("A sunset", model="dall-e-3")              # → LiteLLM
result = await app.ai_generate_image("A sunset", model="openrouter/google/...")  # → OpenRouter

# Audio generation (TTS)
result = await app.ai_generate_audio("Hello", model="tts-1")                    # → LiteLLM
result = await app.ai_generate_audio("Hello", model="fal-ai/kokoro/...")        # → Fal

# Video generation (NEW)
result = await app.ai_generate_video("A cat", model="fal-ai/minimax-video/...")  # → Fal

# Audio transcription (NEW)
result = await app.ai_transcribe_audio(url, model="fal-ai/whisper")             # → Fal

Changes

types.py

• Added fal_api_key field to AIConfig for Fal.ai API key
• Added video_model field with default fal-ai/minimax-video/image-to-video

agent_ai.py

• Added _fal_provider lazy property for cached FalProvider instance
• Updated ai_with_vision() to route fal-ai/ and fal/ prefixed models to FalProvider
• Updated ai_with_audio() to route Fal TTS models to FalProvider
• Added ai_generate_video() method for video generation
• Added ai_transcribe_audio() method for speech-to-text
• Updated docstrings with Fal examples

media_providers.py (from previous commits)

• Added FalProvider with full implementation:
  • generate_image() - Flux, SDXL, Recraft models
  • generate_audio() - Fal TTS models
  • generate_video() - MiniMax, Kling, Luma models
  • transcribe_audio() - Whisper, Wizper models

Tests

• Added comprehensive test suite in tests/test_media_providers.py (30 tests)
• Tests for AIConfig new fields, provider routing, new methods

Provider Routing Summary

Model Prefix	Provider	Modalities
fal-ai/, fal/	FalProvider	image, audio, video, transcribe
openrouter/	OpenRouterProvider	image
(default)	LiteLLMProvider	image, audio (TTS)

Testing

• tests/test_media_providers.py - 30 tests passing
• Syntax validation and import checks

Checklist

• I updated documentation where applicable.
• I added or updated tests (30 new tests in test_media_providers.py).
• I updated CHANGELOG.md (or this change does not warrant a changelog entry).

🤖 Generated with Claude Code

[github-actions]

Performance

SDK	Memory	Δ	Latency	Δ	Tests	Status
Python	9.0 KB	-	0.47 µs	+34%	✓	⚠

✓ No regressions detected