Voice AI agent pricing calculator

Prompt

Create a comprehensive voice AI pricing calculator website with the following features: **Main Calculator Section:** - Header: "VOICE AGENT PRICING CALCULATOR" with subtitle "Calculate the total cost and latency of running voice AI conversations" - Three dropdown selectors for: - LLM Provider (GPT-4o OpenAI, Claude, etc.) - STT Provider (GPT-4o-Transcribe OpenAI, Deepgram, etc.) - TTS Provider (Sonic English Cartesia, ElevenLabs, etc.) **Input Parameters Panel:** - Conversation Length (15 min default) - Agents Per vCPU (1 default) - Words Per Minute (100 default) - Tokens Per Word (1.3 default) - Characters/Word (6 default) - Turns Per Minute (4 default) - LLM Speech Ratio (0.5 default) **Current Pricing Display:** - Transcription Cost/Min: $0.006 - LLM Input Token Cost: $0.0000025000 - LLM Output Token Cost: $0.0000100000 - Voice Cost/Character: $0.000038 - vCPU Cost/Minute: $0.00007083333 **Results Section:** - Large total cost display: "$0.4205" - Cost per minute: "$0.0280" - Warning: "Final prices might be higher!" - Breakdown pie chart showing: Transcription ($0.0900), LLM ($0.1584), Voice ($0.1710), Hosting ($0.0011) - Token counts: Input Tokens (59,475), Output Tokens (975) - Export CSV and social share buttons Make it responsive, modern design with clean typography and intuitive UX. Add a comprehensive latency breakdown section to the voice AI calculator: **Latency Visualization:** - Total Voice-to-Voice Latency: 993ms with color-coded performance indicator - Fast (<800ms) - Green - Acceptable (800-1000ms) - Yellow - Slow (>1000ms) - Red **Three-Stage Breakdown with Progress Bars:** **Input Path (114ms):** - Mic Input: 40ms - Opus Encoding: 21ms - Network Transit: 10ms - Packet Handling: 2ms - Jitter Buffer: 40ms - Opus Decoding: 1ms **AI Processing (790ms):** - Transcription & Endpointing: 300ms - LLM Inference: 350ms - Sentence Aggregation: 20ms - Text-to-Speech: 120ms **Output Path (89ms):** - Opus Encoding: 21ms - Packet Handling: 2ms - Network Transit: 10ms - Jitter Buffer: 40ms - Opus Decoding: 1ms - Speaker Output: 15ms Include explanatory text about voice-to-voice latency and its importance for conversational AI. Add comprehensive formula documentation explaining all calculations: **FORMULA DOCUMENTATION Section:** **Cost Formulas:** - LLM Input Tokens: (words_per_min * tokens_per_word / turns_per_min) * (turns_per_min * convo_length) * (turns_per_min * convo_length + 1) / 2 - LLM Output Tokens: words_per_min * tokens_per_word * llm_speech_ratio * convo_length - Transcription Cost: transcription_cost_per_min * convo_length - LLM Cost: (input_tokens * input_cost) + (output_tokens * output_cost) - Voice Cost: voice_cost * words_per_min * chars_per_word * (1 - llm_speech_ratio) * convo_length - Hosting Cost: (vcpu_cost * convo_length) / agents_per_vcpu - Total Cost: transcription_cost + llm_cost + voice_cost + hosting_cost **Latency Formula:** Total Voice-to-Voice Latency = mic_input + opus_encoding + network + packet_handling + jitter_buffer + opus_decoding + transcription + llm_ttfb + sentence_aggregation + tts_ttfb + opus_encoding + packet_handling + network + jitter_buffer + opus_decoding + speaker_output Include clear explanations for each formula component and real-time calculation updates.

A system prompt was added to support web rendering