Spaces of the week
PII Leak Scanner
Paste output or stream completions to detect personal-data leakage in real time.
Adversarial Red-Team Studio
Interactive jailbreak and prompt-injection playground for any registered model.
RAG Grounding Evaluator
Score answer faithfulness and citation accuracy against your own corpus.
Bias Probe Bench
Run BBQ and BOLD-style probes, get a fairness report card per dimension.
All validation apps, trending first
PII Leak Scanner
Paste output or stream completions to detect personal-data leakage in real time.
EU AI Act Checklist
Interactive obligation mapper for your model's risk classification.
Adversarial Red-Team Studio
Interactive jailbreak and prompt-injection playground for any registered model.
RAG Grounding Evaluator
Score answer faithfulness and citation accuracy against your own corpus.
Multi-Model Judge
Blind pairwise comparison harness with calibrated judge prompts.
Latency SLO Simulator
Replay the probe corpus and visualize p95 breaches under load.
Hallucination Heatmap
Segment responses by claim and color unsupported spans.
Evidence Pack Builder
Assemble audit-ready evidence bundles from completed gate runs.
Guardrail Policy IDE
Author and dry-run policy rules against real production samples.
SOC2 Evidence Mixer
Map validation artifacts onto SOC2 control IDs automatically.
Prompt Injection Arena
Head-to-head injection challenges ranked by defense rate.
Multilingual Toxicity Scan
Toxicity probes across 17 languages with per-language breakdown.
Fairness Report Card
Generate stakeholder-readable fairness summaries from gate runs.
Cost Projection Lab
Estimate monthly token spend across candidate models and providers.
Regression Diff Studio
Visual diff of benchmark scores between two model versions.
Drift Visualizer
Watch distribution shift accumulate across production traffic windows.
Bias Probe Bench
Run BBQ and BOLD-style probes, get a fairness report card per dimension.
Agent Trace Replay
Replay recorded tool-call trajectories and diff expected vs actual actions.
Canary Deploy Watch
Live gate verdicts for canary traffic slices before promotion.
Image Safety Gate
Vision-model caption screening against the image-caption-bias pack.
Token Budget Optimizer
Find the cheapest prompt variant that keeps eval scores above threshold.
GGUF Conversion Gate
Quantization checks for GGUF exports before registry promotion.
Voice Guardrail Tester
ASR stress suite with accented and noisy speech scenarios.
Streaming Speech Judge
Evaluate streaming TTS quality with interruption and latency probes.
Full benchmarks, trade-off lenses, and audit trails are available in the workspace.
Create a free workspace