TOOLS

Design a macro-scale Mixture-of-Experts pipeline from independent local models. Router + domain experts + conditional synthesizer. Generates Python code and YAML config.

Beyond basic VRAM calculator: Input your exact setup (e.g., 2× RTX 4090 + 128GB RAM, or 1× 5090 + CPU offload) and get realistic estimates

Abliteration test suite with publishable results score-card

Generate code to run multiple models in a variety of ways

Test prompt effectiveness

Visual representation for Quant quality V baseline

True VRAM Calculator based on all metrics

Select your GPU — see every uncensored and abliterated model that fits with estimated tok/s and HuggingFace links.

Quick GPU check for uncensored models. Embeddable iframe for Discord and websites.

Predict tokens per second before downloading. 30+ GPUs, all quants, 6 backends.

Detailed profile: throughput, time-to-first-token, bandwidth utilisation, CPU offload analysis. Compare two configs side-by-side.

Current GPU rankings by inference value. Tok/s per dollar, sortable by metric. Updated weekly.

Side-by-side GPU comparison across 7B, 13B, and 70B model sizes with visual bar charts.

4-question wizard giving specific GPU and build recommendations. Budget-aware, use-case aware.

Exact VRAM for any model size, quant, and context length including KV cache breakdown.

Find your maximum context window given VRAM, model, and KV cache quantization.

Plan context usage, generation time, and API cost. Works for local and cloud models.

Answer 3 questions — get the right quantization format with a clear explanation.

4 questions to find the right inference backend for your GPU, OS, and use case.

Compare base vs abliterated benchmark scores. Grade retention S–D. 14 known models.

Side-by-side comparison of base vs abliterated outputs with real examples.

Curated list of abliterated, uncensored, and Dolphin uploads. Updated weekly.

Build a Modelfile with system prompt, sampling params, and context. 6 presets.

20 production-ready system prompts — uncensored assistant, coding, creative writing, reasoning, productivity.

Structured reviews with hardware, use case, verdict, pros and cons.

Share your inference results. Community table with filter and sort.