latency-benchmark

Open Voice AI cost, SIP, realtime pricing, latency benchmark, country deployment, Asterisk, OSS framework, and local call diagnostic tools.

Cost governance for DeepSeek Harness: aggregated token/cost metering per model, session and day, budget caps with threshold alerts and over-limit policies, carbon footprint estimation, per-model latency benchmarks, a Settings budget tab, and the /budget command

AUVBrain: low-latency autonomous AUV control loop with safety monitoring, telemetry, and offline-capable LLM policy support (Ollama/OpenAI-compatible/llama.cpp) plus reproducible latency benchmark proof.

Reproducible GPT Image 2, Flux, Gemini and Nano Banana image API benchmarks with raw outputs, prompt fixtures, latency and cost data.

Reproducible research lab for browser-local small-model RAG with WebGPU/WebLLM runtimes, evidence-packet compression, latency benchmarks, and rights-aware archive fixtures.

ESP32 benchmarks: Polled I/O vs. FreeRTOS ISR context switching latency

Reproducible latency benchmark of scout-postgres vs Laravel Scout's database driver on Postgres 18.

Reproducible OpenAI-compatible API provider benchmark for success rate, P50/P95 latency, response shape and multi-model gateway reliability.

Local-STT voice agent: faster-whisper → tool-calling GPT-4o-mini → macOS TTS, with a reproducible self-contained latency benchmark (3.18s p50 e2e on M4, per-stage receipts). Every number a committed receipt.

Reproducible OpenAI-compatible AI API gateway benchmark with API NODE defaults, latency metrics, success rate, and GitHub Actions results.

可复现的 AI API 中转站测速工具：测试 OpenAI 兼容接口成功率、P50/P95 延迟、错误率和稳定性。

Reproducible end-to-end first-token latency benchmark across the 6 leading AI interview copilots. Raw data + harness.

Multi-Threaded Proxy Health Diagnostic Suite & Anonymity Level Evaluator (Transparent, Anonymous, Elite)

A safe, resumable multi-task benchmark for Hugging Face inference across text, vision, and audio workloads.

Reproducible crypto-trading benchmark data: exchange API latency (p50/p95/p99), real cost-to-trade (fee + spread + slippage), and x402 blockchain settlement latency. Updated automatically. CC BY 4.0. Behind fillbench.com