A sophisticated Discord bot powered by 13 AI models across 5 providers, with intelligent music search, natural language commands, and multi-API integration. Fully deployed and running on Koyeb with real-time uptime monitoring.
| 🧠 13 AI Models Cerebras, SambaNova, Groq, Mistral, Google |
🎵 Music APIs Juice WRLD, Genius, SoundCloud |
💬 Natural Language Conversational commands (no slash required) |
| 🎯 Smart Routing Auto-selects best model for the task |
🎙️ Audio Embed SoundCloud players in Discord |
🔐 Secure Rate limiting, validation, permissions |
- gpt-oss-120b — Best for technical/code analysis
- gpt-oss-120b — General purpose
- mixtral-8x7b-32768 — Fast general purpose
- llama-3.1-8b — Ultra-fast, simple queries
- llama-3.1-70b — Coding & reasoning
- llama-3.1-405b — Complex reasoning & math
- mistral-small — Fast & creative
- mistral-medium — Balanced general purpose
- mistral-large-2411 — Advanced reasoning
- gemini-2.0-flash — Multimodal, web search, fastest
- gemini-1.5-pro — Best reasoning & multimodal
- gemini-1.5-flash — Cost-effective multimodal
- Search all Juice WRLD songs
- Get song details, features, release info
- Full discography browsing
- Featured artist lookup
- Lyrics for any song (non-Juice WRLD)
- Lyric annotations & explanations
- Artist information & biographies
- Music metadata
- Track search with Discord embed player
- Artist profiles with playable tracks
- Playlist discovery with embeds
- Direct playable audio in Discord 🎙️
Just mention the bot naturally:
@bot search for Lucid Dreams
@bot show me the lyrics
@bot what does this lyric mean?
@bot embed this on SoundCloud
@bot Juice WRLD's discography
@bot tell me about the artist
@bot clear my memory
@bot what do you remember?
/status - Bot uptime, current model, health
/models - List all 13 available models
/model <name> - Switch to specific model
/song <query> - Search songs
/lyrics <song> - Get lyrics with annotations
/artist <name> - Artist information
/discography <artist> - Full discography
/features <artist> - Songs with featured artist
/genius <query> - Genius API search
/soundcloud <query> - SoundCloud with embed player
/forget - Clear conversation memory
/stats - Conversation statistics
- 🧠 Automatic context tracking per thread
- 💬 Natural commands: “clear my memory”, “what do you remember?”
- ✅/❌ Reaction-based confirmation for destructive actions
- 📝 Auto-deletion sync: edits/deletes update memory
YamiBot automatically selects the best model based on what you ask:
- Coding/Technical → Cerebras GPT-OSS 120B (or Llama 405B)
- Web Search → Gemini 2.0 Flash (has search built-in)
- Complex Reasoning → Gemini 1.5 Pro or Llama 405B
- Creative Writing → Mistral Large
- Fast Response → Mixtral 8x7B or Mistral Small
- Music Queries → Gemini 2.0 Flash (optimized for music)
- Math/Logic → Llama 3.1 405B
- General Chat → Mixtral 8x7B (balanced)
Manual override available:
@bot use [model_name] for this
When you mention Juice WRLD:
- → Automatically uses Juice WRLD API (primary source)
- → Falls back to Genius only if explicitly requested
When you ask for lyrics (non-Juice):
- → Uses Genius API for comprehensive lyrics + annotations
When you want to embed audio:
- → Uses SoundCloud API with playable Discord player
Example
@bot Juice WRLD songs→ Juice WRLD API@bot show me Humble lyrics→ Genius API@bot embed this on SoundCloud→ SoundCloud with player
- Injection attack prevention
- Spam detection (special characters, repeated chars)
- Message length enforcement (Discord limits)
- @everyone/@here blocking
- Per-user limits: 5 requests/min, 30/hour
- Trusted user multiplier (2×)
- Provider rate limit tracking
- Circuit breaker pattern for API failures
- Memory leak prevention (50MB threshold warnings)
- Automatic conversation cleanup
- Connection pooling & keepalive
- Graceful shutdown handlers
- Admin user support
- Trusted user tiers
- Whitelist/blacklist modes
- Permission levels per user
| Feature | Status | Details |
|---|---|---|
| AI Conversation | ✅ | @mention support, context-aware |
| 13 AI Models | ✅ | All major providers + Google |
| Smart Model Routing | ✅ | Auto-selects best model by intent |
| Natural Language Commands | ✅ | No slash commands needed |
| Slash Commands | ✅ | 12+ commands available |
| Juice WRLD API | ✅ | Full song/artist database |
| Genius API | ✅ | Lyrics + annotations |
| SoundCloud API | ✅ | Embeddable players in Discord |
| Memory Management | ✅ | Thread-aware, auto-cleanup |
| Web Search | ✅ | Via Gemini integration |
| Image Analysis | ✅ | Via Gemini multimodal |
| Rate Limiting | ✅ | Per-user + per-provider |
| Input Validation | ✅ | Injection & spam protection |
| Health Monitoring | ✅ | Provider health checks |
| Circuit Breaker | ✅ | Failure prevention |
| Model Analytics | ✅ | Usage tracking per model |
| User Preferences | ✅ | Remember user settings |
| Reaction Confirmations | ✅ | For destructive actions |
| Message Tracking | ✅ | Edit/delete sync |
| 24/7 Uptime | ✅ | Running on Koyeb |
- Multi-Provider Fallback: Intelligent failover across 5 AI providers
- Async/Await: Full async implementation for performance
- Connection Pooling: Optimized HTTP session management
- Circuit Breakers: Prevents cascading failures
- Model Registry: Centralized model management (13 models)
- Intent Detection: NLP-based command classification
- Response Formatting: Discord-optimized output
- Logging: Comprehensive logging to console + files
- Memory Safe: Proper resource cleanup, no leaks
- AI Models: 13 across 5 providers
- Music APIs: 3 (Juice WRLD, Genius, SoundCloud)
- Commands: 12+ slash commands
- Natural Intents: 13+ recognized intents
- Rate Limits: Per-user + per-provider
- Memory: Automatic leak prevention
- Uptime: 24/7 on Koyeb with monitoring
- Code Quality: Production-grade (type hints, async, error handling)
@bot search for Lucid Dreams
@bot find all Juice WRLD songs with Drake
@bot XXXTentacion's discography
@bot show me Humble lyrics (Genius)
@bot embed songs on SoundCloud
@bot explain quantum computing
@bot help me debug this code
@bot write me a poem
@bot use llama-3.1-405b to solve this math problem
@bot what do you remember about me?
@bot clear my memory
@bot remember that I prefer concise responses
@bot use Gemini for all my future searches
/status - Check bot health
/models - See all available AI models
/stats - Conversation statistics
Status: ✅ Live on Koyeb with uptimebot monitoring
Type: Personal AI Discord Bot
Architecture: Discord.py + FastAPI + Multiple LLM APIs
Features: 60+ integrated capabilities
Last Updated: 2026-01-23
Built for advanced AI conversations, music discovery, and intelligent command routing.