Statsig
Feature flags and experimentation for shipping AI safely.
Acquired — still operating
acquired by OpenAI Sep 2025 (~$1.1B)
About
Statsig is the feature-flag and experimentation platform that AI teams use to ship models without drama. Feature gates swap models or prompts without a deploy, experiments compare model versions head to head, and the stats engine (automatic variance reduction, guardrail metric alerts) keeps results trustworthy.
Its AI SDKs for Python and Node add prompt iteration, evals, and LLM observability hooks, and the AI vertical ties offline evals to online A/B tests. OpenAI acquired Statsig in September 2025 for about $1.1B: a strong signal, and for some teams a vendor-concentration question.
What it can do
- Feature flags with targeting and rollouts
- A/B and multivariate experimentation
- AI-specific evals and observability hooks
- Product analytics and session replay
- Warehouse-native deployment option
- 30+ SDKs and integrations
Setup
- Create an account at statsig.com.
- Install the server or client SDK for your stack.
- Create feature gates and experiments in the console.
- Log exposures and events, then read results.
Prerequisites
A Statsig account; SDK available for 30+ platforms.
Use cases
Works with
Trust score — 8/10
Scored 8/10 (metrics tier, metrics-v1). Evidence: acquired by OpenAI Sept 2025 for ~$1.1B; 1T+ events/day processed; AI SDKs for Python/Node (beta); closed SaaS; SDKs ISC licensed. License: Proprietary SaaS.
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 9.0/10
- Maintenance · recency, releases, issue load
- 9.0/10
- Openness · license + docs
- 3.0/10
- Manual review · sanity read + red-flag check
- 7.0/10
- · acquired by OpenAI Sept 2025 for ~$1.1B
- · 1T+ events/day processed
- · AI SDKs for Python/Node (beta)
- · closed SaaS; SDKs ISC licensed
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.