Results for "budgeting"

Python library that costs and tracks LLM agent runs across models and frameworks, reporting tokens, dollars, and cost per solved task, with LiteLLM and LangChain/LangGraph integrations. MIT licensed.

CI harness for tool-using agents that records tool calls once, replays them in CI, and fails PRs on tool-order or JSON-schema contract violations, tool-call budget overruns, or baseline regressions, with JUnit and HTML reports. MIT licensed.

Self-hosted Apache-2.0 AI gateway and MCP server: per-key virtual tokens with token budgets, exact and lexical-approximate response caching (semantic-grade matching requires an attached embedding endpoint), an append-only audit chain, deterministic local prompt enhancement that makes no provider call, and reverse MCP governance that exposes upstream MCP servers and OpenAPI 3 operations as allow-listed tools.

MCP server and Python library that gives AI agents long-term memory with no LLM call, storing messages word for word in local SQLite or Postgres and recalling what answers a question within a token budget. MIT licensed.

Local-first MCP server for explicit-budget context selection, content-addressed exact recovery, and auditable Context Receipts. Apache 2.0 licensed.

LLM gateway with a unified OpenAI-compatible API across providers, routing, load balancing, fallbacks, guardrails, and observability. Apache 2.0 licensed.

Production-grade multi-agent orchestration framework with 10K+ stars. Apache 2.0 licensed.

AI agent company and orchestration framework with 55K+ stars. MIT licensed.

Local-first control plane for routing coding work across Claude Code, Codex, Cursor, and OpenCode, with quota-aware rotation across multiple Claude/Codex subscription profiles, shared thread context, best-of-N runs, and cross-model review.

Self-hosted platform for running AI employees for small businesses, each with its own mailbox, job description, governed procedures, memory, and place on an org chart, with observable desktop, browser, and terminal sessions. AGPL-3.0 licensed.

Production SDK and harness runtime for building, monitoring, and controlling end-to-end AI agent lifecycles in Python and TypeScript. Apache 2.0 licensed.

Local-first desktop multi-agent harness with persistent workspaces, capability-scoped tools, agent memory, budgets, schedules, and live runtime visualization. MIT licensed.

TypeScript framework for building reliable AI applications. "Official" DSPy-inspired framework for TypeScript with type-safe LLM interactions, chain-of-thought reasoning, and structured output validation. Apache 2.0 licensed.

Ultra-fast custom kernels for training speedup.

Official library with LoRA, QLoRA, DoRA, etc.

Global weather and tropical cyclone forecasting framework from Google DeepMind, including WeatherNext 2, GraphCast, and GenCast models. Apache 2.0 licensed.

High-accuracy open-source biomolecular structure prediction model from ByteDance. First fully open-source model to outperform AlphaFold3 across diverse benchmarks with Apache 2.0 licensing for both academic and commercial use.

Open-source financial large language models. Democratizing financial AI with data-centric training pipeline and multiple model releases for trading, analysis, and robo-advising. MIT licensed.

PyTorch runtime and toolchain for deploying AI models on mobile, embedded, and edge devices.

Self-hosted AI workspace combining agents, skills, and tools (Gmail, Calendar, file conversion) that runs entirely on your own provider API keys, sealed in the browser so the server stores none of them. Apache-2.0 licensed.

CLI that reads a project's specification files and recommends which LLM fits the work, using a local judge model via Ollama and citing the spec fragment behind each demand level. MIT licensed.

Local structural code search, caller navigation, impact analysis, and entity-level diffs exposed to coding agents through a CLI and MCP server. Apache-2.0 licensed.

24/7 AI employee team for automating complex challenges. Low-code multi-agent framework with handoffs, guardrails, memory, RAG, and 100+ LLM providers.

Unified AI gateway and proxy supporting over 230 providers with token compression, automatic failover, and routing strategies. MIT licensed.