awesome-repositories.com
Blog
MCP
awesome-repositories.com

Discover the best open-source repositories with AI-powered search.

ExploreCurated searchesOpen-source alternativesSelf-hosted softwareBlogSitemap
ProjectMCP serverAboutHow we rankPress
LegalPrivacyTerms
© 2026 Bringes Technology SRL·VAT RO45896025·hello@awesome-repositories.com
experientiallabs avatar

experientiallabs/experiential

0
View on GitHub↗

Experiential

Experiential is an open-source LLM gateway and router. It exposes OpenAI-compatible and Anthropic-compatible APIs in front of hosted models, models reached with your own provider keys, and local models, all under one base URL and one bearer key. Existing OpenAI and Anthropic SDK clients, as well as coding agents, can be pointed at it by changing only the endpoint. The project states that usage carries zero markup.

What sets it apart is that routing is learned from your own traffic. It captures traces from desktop coding agents and imports past activity from supported monitoring tools, databases, or exported trace files, keeping prompts, tools, results, and model identity as reusable evidence. From that evidence it builds representative task sets, simulates candidate models, locks a routing policy, and verifies it on held-out tasks within fixed spend limits, with the aim of cutting cost while preserving quality. The same traces can also be used to fine-tune an open-source model that you own, and agent episodes can be run against a candidate model in local or remote sandboxes.

Around the gateway is a governance layer that controls which users and agents may call which models and enforces spending limits across a team. Free access can be capped per model by requests and tokens, with an option to continue by spending a credit balance at list price. Accounts and API keys can be created from just an email address in a single request. Locally, the gateway serves chat, responses, and model-list routes, routing by alias or saved policy, with either durable logging or a no-save mode. It also falls back to a conservative baseline if request-time embedding fails and marks aliases that cannot be built as unavailable while the rest keep working.

AI search

Explore more awesome repositories

Describe what you need in plain English — the AI ranks thousands of curated open-source projects by relevance.

Start searching with AI
experientiallabs.ai
↗

Features

  • LLM Gateways - Open-source gateway exposing OpenAI- and Anthropic-compatible APIs in front of hosted, own-key, and local models with zero markup.
  • Trace-Learned Router Builders - Builds a custom router tuned for quality, speed, and cost by learning from collected traces of an existing agent's production traffic.
  • LLM Gateways - Lets existing OpenAI and Anthropic clients and coding agents reach hosted, own-key, and local models through one gateway endpoint.
  • OpenAI-Compatible APIs - Serves chat completions and responses under one base URL and bearer key so existing OpenAI client code works by changing only the endpoint.
  • Anthropic-Compatible Endpoints - Accepts Anthropic Messages requests on the same gateway and key so Anthropic SDK clients work by changing only the base URL.
  • OpenAI-Compatible Endpoints - Serves chat completions, responses, and model lists under one base URL and bearer key so OpenAI-style clients work unchanged.
  • Model Request Routing - Puts one OpenAI-compatible API in front of hosted, bring-your-own-key, and local models so applications can switch or mix providers without code changes.
  • Cost-Aware Model Routers - Simulates candidate models, locks a routing policy, and verifies it on held-out tasks within fixed spend limits to cut cost while keeping quality.
  • LLM Traffic Trace Collection - Captures coding agent traffic and imports past activity from monitoring tools, databases, or exported files as reusable trace evidence.
  • Cost-Quality Model Routing - Learns from real traffic to route each request to a cheaper model that still meets quality needs, verified on held-out tasks.
  • Trace-Driven Router Optimizers - Learns from captured agent traffic to simulate candidate models, lock a routing policy, and verify it on held-out tasks.
  • Coding Agent Traffic Capture - Records traces from desktop coding agents on existing subscriptions while requests go straight to the provider, and reports what was captured.
  • Agent Gateways - Lets existing coding agents send requests through the gateway by setting only a base URL and key in the environment.
  • Bring-Your-Own-Key Model Access - Connects users' own inference provider keys so model calls run on their accounts while the gateway lists models and controls access.
  • Coding Agent Traffic Capture - Records traces from desktop coding agents using existing subscriptions while requests go straight to the provider.
  • Custom Model Training - Fine-tunes an open-source model that the user owns from collected traces and task sets, tuned to their own workload.
  • OpenAI-Compatible Model Servers - Runs a local server with chat, responses, and model-list routes, routing by alias or saved policy with durable logging or a no-save mode.
  • Representative Task Set Builders - Builds training and held-out task sets, search indexes, and a grounded environment model from saved trace evidence for review before tuning.
  • Model Fine-Tuning - Fine-tunes an open-source model that the user owns from collected routing traces to tune it to their own workload.
  • Free Usage Caps with Paid Continuation - Caps free access per model by requests and tokens and lets users keep going by spending a credit balance at list price.
  • Trace Data Importers - Parses and normalizes agent traces from files or captured traffic into shared storage, keeping prompts, tools, results, and model identity, with repeat-safe imports.
  • Managed Agent Sandboxes - Runs full agent episodes against a candidate model inside local or remote sandboxes with limited processes, metered usage, and verified cleanup.
  • AI Token Spend Limits - Enforces spending limits on model usage and caps free usage by requests and tokens, with optional paid continuation at list price.
  • User-Provided Keys - Finds a user's existing provider API keys, confirms each with the user, and connects approved ones so calls run on the user's own accounts.
  • Model Access Governance - Provides a governance layer controlling which users and agents may call which models and enforcing team-wide spending limits.
  • AI Provider Routing - Routes requests across hosted, bring-your-own-key, and local model providers behind a single OpenAI-compatible endpoint.
15,193 stars·292 forks·Python·Apache-2.0·6 views

Star history

Star history chart for experientiallabs/experientialStar history chart for experientiallabs/experiential

How this analysis was created: This summary and feature list are AI-generated from collected project material and can contain mistakes. Stars, license and language are imported from GitHub. Inclusion does not mean that we have tested or audited this project. Check the source documentation for any feature you depend on. Learn more on our About page.

Frequently asked questions

What does experientiallabs/experiential do?

Experiential is an open-source LLM gateway and router. It exposes OpenAI-compatible and Anthropic-compatible APIs in front of hosted models, models reached with your own provider keys, and local models, all under one base URL and one bearer key. Existing OpenAI and Anthropic SDK clients, as well as coding agents, can be pointed at it by changing only the endpoint. The project states that usage carries zero markup.

What are the main features of experientiallabs/experiential?

The main features of experientiallabs/experiential are: LLM Gateways, Trace-Learned Router Builders, OpenAI-Compatible APIs, Anthropic-Compatible Endpoints, OpenAI-Compatible Endpoints, Model Request Routing, Cost-Aware Model Routers, LLM Traffic Trace Collection.

Which projects share features with experientiallabs/experiential?

Projects with overlapping indexed features include: nvidia-nemo/switchyard — Switchyard is a Rust library for routing large language model requests. For each request it chooses which model and… helicone/helicone — Helicone is an AI gateway and observability platform designed to intercept, manage, and monitor interactions with… mnfst/llm-gateway — LLM Gateway is an open-source, self-hostable gateway that sits between applications and AI model providers. It accepts… lm-sys/routellm — RouteLLM is a routing framework and traffic manager designed to direct prompts between high-capability and low-cost… tashfeenahmed/freellmapi — freellmapi is a self-hosted gateway that combines the free tiers of many language model providers behind one local,… xiangsx/gpt4free-ts — gpt4free-ts is a TypeScript-based LLM API proxy and gateway that provides a unified interface for accessing large…

Projects sharing features with Experiential

These projects share indexed features with Experiential. Shared tags can include platform or build tooling; verify the primary use case before treating a result as a replacement.
  • nvidia-nemo/switchyardNVIDIA-NeMo avatar

    NVIDIA-NeMo/Switchyard

    3,494View on GitHub↗

    Switchyard is a Rust library for routing large language model requests. For each request it chooses which model and provider should respond, balancing answer quality, cost, and response time rather than relying on one fixed model. It is aimed at coding assistants and agent harnesses, where it picks the best available model for each step and can show the chosen model in a tool's status line. Routing decisions come from composable strategies. A language model can judge whether a cheaper, more efficient model can handle a task, and that verdict selects the model tier. Execution signals, such as

    Rust
    View on GitHub↗3,494
  • helicone/heliconeHelicone avatar

    Helicone/helicone

    6,273View on GitHub↗

    Helicone is an AI gateway and observability platform designed to intercept, manage, and monitor interactions with large language models. By acting as a reverse-proxy, it provides a centralized layer for routing requests across multiple AI providers, allowing developers to maintain consistent application logic while gaining deep visibility into model performance, usage, and costs. The platform distinguishes itself through a robust suite of traffic management and prompt engineering tools. It enables policy-driven control, including automatic failover between providers, rate limiting, and edge-b

    TypeScript
    View on GitHub↗6,273
  • mnfst/llm-gatewaymnfst avatar

    mnfst/llm-gateway

    7,625View on GitHub↗

    LLM Gateway is an open-source, self-hostable gateway that sits between applications and AI model providers. It accepts requests at a single OpenAI-compatible chat completions endpoint, so any SDK that allows a custom base URL can send traffic through it, and forwards each request to a suitable model from the connected providers. Each agent uses its own key, and a PostgreSQL database backs the service. What sets it apart is the routing and cost layer. It can score each request and send it to the cheapest model able to handle it, pick a model automatically when the model is set to auto, or appl

    TypeScriptaiai-ai-sdk
    View on GitHub↗7,625
  • lm-sys/routellmlm-sys avatar

    lm-sys/RouteLLM

    5,080View on GitHub↗

    RouteLLM is a routing framework and traffic manager designed to direct prompts between high-capability and low-cost large language models. It functions as an API gateway that mimics the OpenAI specification to route requests across different model providers. The system optimizes operational costs by splitting traffic between model tiers based on predicted win rates and prompt complexity. It includes a calibration tool to analyze sample queries and determine the optimal cost-quality tradeoff for traffic distribution. The framework provides a tool for measuring the accuracy and cost efficiency

    Python
    View on GitHub↗5,080
Compare all 30 related projects→