Cohesor is an AI gateway and control plane for agents that routes requests, compresses tokens, brokers MCP tool calls, and enforces policy from one endpoint. It is designed for teams that want to lower model spend and improve visibility without rewriting existing Anthropic- or OpenAI-compatible clients.
Cohesor is an AI gateway and control plane for agent traffic. It sits between agents and model providers to route requests, compress tokens, broker MCP tool calls, and enforce policy before requests reach a model or external tool.
The product is aimed at teams running AI agents that want one endpoint for multiple models and tools, plus controls for spend and visibility. The site describes Anthropic and OpenAI protocol compatibility, support for common agent clients, and a usage-based pricing model built around a per-request platform fee plus provider-cost LLM usage.
Point agents to a single Cohesor URL and use Anthropic- and OpenAI-compatible clients without rewriting the agent. The product says Claude Code, Codex, opencode, and any compatible SDK can connect with a one-line configuration change.
A lightweight complexity classifier scores each request and sends it to either a stronger model or a lower-cost model. The router can be tuned by sensitivity and can also be pinned to a fixed model when routing is disabled.
The gateway compresses prompts, conversation history, and tool output before billing. Cohesor describes this as lossless-first and says the transforms run inline without extra round trips.
Cohesor exposes an audited MCP endpoint and brokers OAuth credentials, API keys, and per-agent scopes. Tool calls are traced with arguments, latency, and cost so they can be reviewed later.
The dashboard tracks traces, per-user spend, model routing, tool usage, and audit logs. Pricing and home-page copy also describe team budgets, caps, and showback reporting.
Point Claude Code, Codex, or another compatible client at Cohesor to keep the same workflow while moving traffic through a gateway that can route, compress, and govern requests.
Use request complexity scoring to send trivial or low-risk work to cheaper models while keeping harder prompts on stronger models, with the option to adjust sensitivity or pin a fixed model.
Broker Gmail, Drive, Calendar, GitHub, and other MCP tools through one audited endpoint with per-agent scopes and credential handling outside the prompt.
Track spend by user, team, and API key, set caps, and review traces or audit logs when you need to understand where tokens, tool calls, and dollars went.
Compress prompts, conversation history, and tool outputs before they are billed to reduce input size while keeping the working context intact for the model.
Cohesor is a neutral control plane for AI agents that combines routing, token compression, MCP tool management, governance, and observability behind one endpoint. It is intended for teams building or running agents that want to reduce model spend without changing their existing client setup.
The pricing page says Cohesor uses usage-based pricing: LLM usage is billed at provider cost, plus a flat platform fee of $10 per 100,000 requests on the Pay as you go plan. Enterprise uses custom volume pricing and terms.
No. The pricing and FAQ pages say you can point your agent at Cohesor with one environment variable, and Anthropic/OpenAI-compatible clients such as Claude Code and Codex work unchanged.
Compression is described as lossless-first by default, with meaning preserved rather than every byte. Routing can also be pinned to a fixed model when you do not want automatic lane selection.
Cohesor says provider keys are encrypted at rest, gateway keys are hashed, and MCP OAuth tokens stay client-side. Its MCP page also says credentials are brokered at call time and never exposed in prompts or logs.
AakarDev AI helps teams manage AI provider access, project-level setups, logs, and analytics from one dashboard. It supports BYOK workflows and lists providers including OpenAI, Google Gemini, Anthropic, Groq, Mistral AI, and Perplexity AI.
CreateOS Sandbox is an isolated compute environment for running code and agent workloads inside Firecracker micro-VMs. It is designed for workflows that need machine-level isolation, private networking between sandboxes, and programmatic control through SDK, CLI, or MCP.
Trigger.dev chat agent is a durable AI chat backend for developers building stateful conversations that can survive refreshes, crashes, and long-running turns. It connects with the AI SDK `useChat` flow and runs on managed infrastructure with no timeout on a turn.
Arduino VENTUNO Q is an edge AI computer for AI and robotics applications. It combines AI inference and deterministic control on a single board and is designed to work with Arduino App Lab.
ByteAsk is a terminal-first AI coding agent for C and C++ that edits repositories and verifies changes with the real compiler, debugger, sanitizers, and tests before showing a diff. It offers a free tier plus paid plans, with editor connectors and zero-retention handling described in the source.
Codex Plugins bundle reusable skills, app integrations, and MCP servers into workflows you can install in the Codex app or use from Codex CLI. They help extend Codex with connected-service tasks, reusable instructions, and shared team workflows.
