# Frametail

> Production observability for generative video: traces, benchmarks, and evaluations so teams compare outputs, catch regressions, and ship with evidence.

Frametail helps AI product and engineering teams evaluate AI-generated video outputs with observability built around the clip, not generic text traces alone. Pin datasets, run immutable benchmarks with scorers, read production traces beside artifacts, and iterate in Sandbox before promoting to sign-off.

## Audience

- AI product managers and applied ML engineers shipping generative video
- Teams that need linkable benchmark artifacts for release reviews
- **Beachhead:** Teams building on fal.ai — Frametail is the evaluation layer for traces and benchmarks on top of fal inference

## Key pages

- Home: https://frametail.com/
- Documentation: https://frametail.com/docs
- Blog: https://frametail.com/blog
- Pricing (free tier): https://frametail.com/pricing
- FAQ (structured answers): https://frametail.com/faq
- Book a demo: https://frametail.com/demo
- Learn hub: https://frametail.com/learn
- Glossary: https://frametail.com/learn/glossary
- Integrations: https://frametail.com/integrations
- fal.ai integration: https://frametail.com/integrations/fal
- OpenRouter integration: https://frametail.com/integrations/openrouter
- Langfuse alternative: https://frametail.com/alternatives/langfuse
- Frametail vs Langfuse: https://frametail.com/compare/frametail-vs-langfuse
- For ML engineers: https://frametail.com/for/ml-engineers

## Pricing

**Free** ($0/mo): 500 benchmark rows/mo, 1,000 scorer runs/mo, 14-day trace retention, unlimited members, up to 2 projects, unlimited datasets & sandbox. Book a demo: https://frametail.com/demo

**Pro** ($100/mo): 25k benchmark rows/mo, 25k scorer runs/mo, 30-day retention, live scoring, priority support. Book a demo: https://frametail.com/demo

**Enterprise** (demo-led): custom limits, SSO/SAML, RBAC, custom retention/export, premium support. Book a demo: https://frametail.com/demo or founders@frametail.com

Book a demo to get started: https://frametail.com/demo

See also: https://frametail.com/pricing.md

## Product capabilities

- Immutable benchmarks on pinned datasets with scorer contracts
- Traces with spans beside video artifacts
- Sandbox threads with promote-to-benchmark workflow
- Live scoring on production traces (benchmarks remain explicit)
- Integrations: fal.ai, OpenRouter, custom endpoints via provider-agnostic SDK

## Common questions (extractable answers)

**What is Frametail?**
Frametail is a generative video evaluation platform. Teams run immutable benchmarks on pinned datasets, read production traces beside video artifacts, and iterate in Sandbox before promoting experiments into scored, linkable benchmark runs.

**What is generative video evaluation?**
Generative video evaluation is the practice of comparing AI-generated video outputs across model, prompt, or pipeline changes using pinned datasets, scorer contracts, and reproducible benchmark runs — not ad-hoc clip review in Slack.

**Best tool for evaluating fal.ai video generations?**
Frametail is built as the evaluation layer for teams on fal.ai: automatic tracing via the fal client SDK, benchmarks on pinned datasets, and sandbox-to-benchmark promotion. See https://frametail.com/integrations/fal

**Frametail vs Langfuse for video?**
Langfuse is strong for general LLM tracing; Frametail is purpose-built for generative video — artifact-native traces, immutable benchmarks, and sandbox promotion. Comparison: https://frametail.com/compare/frametail-vs-langfuse

**Category:** Generative video evaluation / observability (not generic text LLM tracing)

**Positioning:** The evaluation layer for teams building on fal.ai.