Agentic coding CLI that works with any OpenAI-compatible LLM — GPT-5.x, DeepSeek, Gemini, Ollama, and 200+ models

Core types and registry utilities for TokenLens (model metadata).

Typed client for models.dev to fetch model catalogs with friendly errors.

OpenCode plugin that enables 1M token context window for Claude and GPT-5 models on any provider (Anthropic direct, third-party gateways, Amazon Bedrock, GitHub Copilot, etc.)

Enhanced GPT-5 MCP core with TypeScript, caching, and advanced features

Official Node.js SDK for Crazyrouter — One API key for 300+ AI models (GPT-5, Claude Opus 4, Gemini 3, DeepSeek V3 and more). Drop-in replacement for OpenAI SDK.

Javascript BPE Encoder Decoder for GPT-2 / GPT-3

Pi extension for image generation via OpenAI gpt-image, Google Nano Banana (Gemini), Alibaba Qwen-Image, OpenRouter, and custom providers.

Fast token estimation with 95%+ average accuracy in a 2kB bundle

Calculate OpenAI API costs before making calls. Supports GPT-4o, GPT-4.1, o3, o4-mini, GPT-5, and more.

Offline BPE tokenizer for OpenAI, Anthropic, and Gemini — zero dependencies

Use web-only models (e.g., GPT-5 Pro) with your local codebase: scan, pack to token limits, and copy a paste-ready bundle.

A lightweight coding agent designed for small LLMs (e.g., GPT-5 mini, MiniMax M2.7, Claude Haiku). Focuses on reliability under model limits, with benchmarks comparing performance against agents like OpenCode.

Calculate the token consumption and amount of openai gpt message

Connects Sling CLI to Publicis Sapient Slingshot enterprise LLM gateway. Bundles the pi coding-agent runtime.

MagnetAPI.org Codex CLI for ChatGPT-6-Astra, Fable 5.1, Opus 5, Sonnet 5, Daybreak Blue, and GPT-5.6.

Pentect provider extension for Pi

LLM provider abstraction layer with unified streaming interface

OpenAI ChatGPT (Codex backend) OAuth auth plugin for opencode - use your ChatGPT Plus/Pro subscription instead of API credits

Helpers for context windows, usage normalization, compaction, and cost estimation.