The Complete LLM Infrastructure & Token Toolkit
Calculate GPU VRAM requirements across 53+ open-weight models (DeepSeek-V4.1, Kimi-K3.1, GLM-5.3, Qwen3.8, Gemma 4), compare 41 frontier API prices against self-hosted GPUs, and inspect agent harness curves.
Core Interactive Tools
LLM GPU Memory Calculator
Calculate weights, KV cache, and activation VRAM across FP16, FP8, MXFP4, INT8 & INT4 plus generate vLLM / Ollama commands.
API Pricing & Break-Even Calculator
Compare monthly API spend for GPT-6.2 Astra/Sol, Claude 5.5, Gemini 4, Qwen4, DeepSeek-V4.1, Kimi-K3.1 & GLM-5.3 with prompt caching and GPU break-even.
Frontier Cost-Effectiveness Curves
Compare 34 frontier LLMs (GPT-6.2 Astra/Sol, Claude Fable 5.2 / 5.5, Gemini 4 Carbon, Qwen4, DeepSeek-V4.1) & 6 coding agent harnesses on Pareto curves.
GPU LLM Compatibility Finder
Pick an RTX 5090, RTX 4090, Apple M4 Max, H100, or A100 to see every open-weight model it can run locally and at what speed.
Token Counter & Cost Estimator
Inspect exact BPE tokens offline across GPT-6, Claude 5.5, Gemini 4, Qwen4/3.8, Kimi-K3.1 & DeepSeek-V4.1 with instant API cost previews.
Token Generation Speed Simulator
Experience real-time 1-200 tokens/sec LLM streaming speed to test UX latency for chat, code assistants, and long-form tools.
Voice AI & Full-Duplex Cost Calculator
Compare OpenAI gpt-live-1 vs gpt-realtime-2.1, Gemini 3.8 Live, Vapi/Retell, self-hosted PersonaPlex-7B & zero-upload audio duration ASR/TTS batch bills.
Model-Specific VRAM & Hardware Guides (85 Pages)
Select any dense or MoE open-weight model for precomputed FP16/FP8/INT4 VRAM matrices and GPU counts.