Folkbench
Uptime, latency and price for AI API relays
Folkbench is a measurement toolkit for AI API gateways and relays. It publishes per-model success rate, response time and listed price, and ships Folkbench Switch, an open-source desktop app that writes a chosen Base URL and key into Claude Code, Codex or Gemini CLI after a diff preview.
Key Features & Capabilities
Core highlights, automation workflows, and capabilities offered by this tool
Standardized relay rankings
Compares services for the same AI model using the same published evaluation criteria, helping users assess availability, latency, pricing, and report status without mixing incompatible benchmarks.
Published evaluation reports
Provides service-level reports that document what was tested, when the check occurred, response timing, availability outcomes, and notes on errors, timeouts, or missing data.
Image and video model evaluation
Tracks media-generation services with signals such as success rate, P95 generation time, published samples, service coverage, and official pricing where available.
On-demand service testing: Lets users select a listed model and service, confirm authorization to test, and receive a report for a more current, decision-specific service check.
AI developers and teams choosing an AI API relay or gatewayWithin-model comparison of AI API relays
How It Works (Step-by-Step)
The end-to-end user journey from input to automated output
Step 1
Browse the directory and select an AI model and API relay/service.
Step 2
Review published evaluation signals, including availability, response speed, pricing, report status, and—where available—media-generation success rate and P95 generation time.
Step 3
Open a service report or run an on-demand check when you are authorized to test the service, then use the resulting report to compare providers.
Frequently Asked Questions
Quick answers to common questions regarding functionality, billing, and access