# Tempr

> Tempr ships three products on one account: Chat, an agent toolset inside Visual Studio, VS Code and JetBrains IDEs; CLI, the same agent in your terminal; and Gateway, an OpenAI-compatible API for your own provider keys. A fourth, Tempr Code - the agent in its own desktop app - is in development. All are bring-your-own-key (BYOK) - Tempr never fronts provider tokens or takes a cut of them.

Key facts, from https://temprhq.io/faq:

- What is Tempr? Tempr is a platform for running AI coding agents and model traffic on your own provider keys. It puts the same agent in your IDE (Chat), your terminal and CI (CLI) and, soon, its own desktop app (Code), and gives your own apps an OpenAI-compatible API (Gateway). All of it shares one account and one Portal, where an organization manages its provider keys, budgets and per-run cost limits, and sees every agent run traced call by call. You pay your providers directly for tokens; Tempr charges for the product, never for the tokens.
- How much does Tempr cost? Chat is Standard $15/mo or Pro $20/mo (Teams from $35/seat/mo), the CLI is included with every Chat plan, and the Gateway starts free with paid tiers at $49/mo and $249/mo, which include one and five Chat and CLI seats (more at $35/seat/mo for a team). Full breakdowns and comparison tables are on the pricing page.
- Is there a free plan? The Gateway has a free tier - 10,000 requests a month, no card required. Chat and the CLI don't have a free plan, but every paid Chat plan starts with a 7-day free trial, and the CLI is included with it.
- Does Tempr mark up token prices? No. Every request goes out on a key you supplied, at your provider's own rates - Tempr never fronts tokens or takes a cut. What you pay Tempr for is the product wrapped around your key: the IDE extension and CLI tooling, or the Gateway's virtual keys, logs, and analytics.
- Which providers are supported? Around 30 today - first-party providers, inference hosts, cloud platforms, and aggregators, all enforced by a server-side allowlist. See the full, current list on the providers page; if yours isn't listed yet, tell us and we'll look at adding it. You can also run models on your own OpenAI-compatible server as custom models.
- Is my code sent anywhere besides the provider I choose? Your requests are relayed through our servers to the provider or aggregator you configure, always using your own key - never pooled or shared with other customers. When you use agent mode, the multi-turn tool-calling loop runs server-side for reliability (so it can survive a dropped connection and resume mid-turn); tool-call requests and results, which can include snippets of your code, pass through our servers and are retained there to support that run.
- How are my provider keys stored? Encrypted at rest, never logged in plain text, and never pooled or shared across customers - requests always go out on the key you supplied, for the account you configured.

## Product pages

- [Tempr](https://temprhq.io/): AI coding agents for your IDE, terminal and desktop, and an OpenAI-compatible gateway for your apps. On your own provider keys, with budgets, traces and one Portal.
- [Tempr Chat](https://temprhq.io/chat): Bring your own API key to OpenAI, Anthropic, Google & OpenRouter - in-IDE chat, inline completion, and agent tools inside Visual Studio, VS Code, and JetBrains IDEs. 7-day free trial.
- [Tempr CLI](https://temprhq.io/cli): Tempr CLI: the same BYOK agent you use in your IDE, now in the terminal. One-shot prompts, an interactive REPL with persistent history, and piped headless mode for scripts - running on your existing Tempr license.
- [Tempr Gateway](https://temprhq.io/gateway): Tempr Gateway: an OpenAI-compatible API for your own provider keys and the AI plans you already pay for. Run Claude Code on your Claude plan, Codex on ChatGPT, or any tool on 30 coding plans, with virtual keys, request logs, guardrails and budgets, plus MCP connectors that give claude.ai and ChatGPT your MCP servers.
- [Tempr Code](https://temprhq.io/code): Tempr Code is a bring-your-own-key coding agent in its own desktop app for Windows, macOS, and Linux - parallel chats on isolated worktrees, a built-in terminal, diff review before anything lands, MCP, and subagents. In development.
- [Pricing](https://temprhq.io/pricing): Tempr pricing: Chat (Standard $15/mo, Pro $20/mo, Teams from $35/seat), the CLI (included with your Chat license), and the Gateway (Free, Pro $49/mo, Scale $249/mo, with Chat and CLI seats included on Pro and Scale). Tempr Code is in development, pricing to be announced. Every plan is bring-your-own-key.
- [Model catalog](https://temprhq.io/models): Every model available through Tempr - 436 models across OpenAI, Anthropic, Google, DeepSeek, Mistral, and more, with reference pricing, context windows, and capabilities. Bring your own key and use them in your IDE, your terminal, or your API.
- [Providers](https://temprhq.io/providers): Every provider on Tempr's supported allowlist - OpenAI, Anthropic, Google Gemini, DeepSeek, Mistral, OpenRouter, and more. Bring your own key in Chat, the CLI, and the Gateway.
- [FAQ](https://temprhq.io/faq): Frequently asked questions about Tempr: the products (Chat, CLI, Gateway, plus Tempr Code in development), bring-your-own-key, privacy, billing, plans, and trials.
- [MCP servers](https://temprhq.io/mcp): Browse the Tempr MCP catalog: curated, one-click-install Model Context Protocol servers that extend the agent with your own tools and data - filesystem, GitHub, Postgres, Playwright, and more.
- [Changelog](https://temprhq.io/changelog): What's new in Tempr: every release of the Tempr Gateway, the VS Code, JetBrains and Visual Studio extensions, the CLI and the Portal, newest first.
- [Blog](https://temprhq.io/blog): Notes from the team building Tempr: how the Gateway, the IDE extensions and the CLI work, and what we learn running them across every AI provider.

## Comparisons

- [Tempr vs GitHub Copilot, Cursor, Cline, and Continue](https://temprhq.io/compare): How Tempr compares with GitHub Copilot, Cursor, Cline, and Continue: IDE support, what your own API key covers, inline completion, pricing, and where your requests go. Checked September 2026.
- [Tempr vs GitHub Copilot](https://temprhq.io/tempr-vs-github-copilot): Tempr vs GitHub Copilot in Visual Studio, VS Code, and JetBrains IDEs: what your own API key covers, inline completion, pricing, and where your code goes. Checked September 2026.
- [Tempr vs Cursor](https://temprhq.io/tempr-vs-cursor): Tempr vs Cursor: an extension for Visual Studio, VS Code, and JetBrains IDEs versus a separate AI editor. What your own API key covers, pricing, and where your code goes. Checked September 2026.
- [Tempr vs Cline](https://temprhq.io/tempr-vs-cline): Tempr vs Cline: IDE support including Visual Studio, inline completion, agent features, pricing, and where your requests go. Both run on your own API keys. Checked September 2026.
- [Tempr vs Continue (discontinued)](https://temprhq.io/tempr-vs-continue): Continue was acquired by Cursor in June 2026 and is no longer maintained. How Tempr compares for bring-your-own-key chat, agent, and autocomplete in VS Code and JetBrains IDEs, and how to move over. Checked September 2026.

## Blog

- [Putting a price ceiling on an agent run](https://temprhq.io/blog/a-price-ceiling-for-agent-runs): An agent run is a loop of model calls whose cost you can't know in advance. How Tempr lets you, and your organization, cap what one run may spend, and see which runs cost the most.
- [A sandbox for the agent's commands, with nothing to install](https://temprhq.io/blog/a-sandbox-with-nothing-to-install): How the Tempr CLI lets an agent run any build or test command without asking, by limiting what those commands can touch. Why we built it on Linux's Landlock instead of the usual tools, and what it took to make real builds work inside it.
- [Budgets that hold when requests run at the same time](https://temprhq.io/blog/budgets-that-hold-under-concurrency): An ordinary budget check lets requests that run together spend past the limit together. How Tempr's Strict option sets aside each request's worst case while it runs, what that costs, and why it's off by default.
- [Every MCP tool call, through one place you control](https://temprhq.io/blog/every-mcp-tool-call-through-one-place): MCP gives coding agents real tools, along with tokens on every laptop, shared bot accounts and tool output the model trusts. How Tempr's Gateway relays MCP traffic so it runs as the right person, gets checked both ways, and leaves a record.
- [One plan, many keys: how limits work across a team's Gateway keys](https://temprhq.io/blog/one-plan-many-keys): A team rarely has one Gateway key. It has one for each app, each CI job and each person. How Tempr's plan limits, key limits, member budgets and organization caps fit together, and why each one sits where it does.
- [One reasoning setting for every provider, and what it took](https://temprhq.io/blog/one-reasoning-setting-every-provider): Every AI lab names its reasoning controls differently, and the same model can behave differently on two hosts. How Tempr gives you one setting that works everywhere, and what we found testing it.
- [OpenAI's newest models, on the API everything else speaks](https://temprhq.io/blog/openai-newest-models-on-chat-completions): OpenAI's pro, codex and newest gpt-6 models do some things, or everything, only on its Responses API. How Tempr bridges chat completions requests to them, one request at a time, so your tools and agents keep working.
- [Paid MCP tools: refused clearly, never paid](https://temprhq.io/blog/paid-mcp-tools-refused-never-paid): Some MCP servers now charge per tool call, through x402 or MPP. What an agent does when a tool asks to be paid, why Tempr doesn't pay, and how it makes the refusal clear instead of a confusing error.
- [Pipe it to the agent: the Tempr CLI in scripts and CI](https://temprhq.io/blog/piping-into-the-tempr-cli): The Tempr CLI reads stdin, writes stdout and exits with a code that means something, so an AI agent fits into a shell pipeline or a CI job like any other command. Patterns that work, and what we fixed so they do.
- [Running Codex CLI on Claude, and every other model](https://temprhq.io/blog/running-codex-cli-on-any-model): Codex CLI speaks OpenAI's Responses API, so out of the box it only talks to OpenAI. How Tempr Gateway's /v1/responses lets it run on Claude, Gemini, DeepSeek and the rest, what carries over, what doesn't, and the fixes it took to make sub-agents and request logs work.
- [Seeing your Gateway traffic: a tour of the Portal](https://temprhq.io/blog/seeing-your-gateway-traffic): The questions a team asks about its model traffic, such as why cost jumped on Tuesday or which customer is spending, and where the Tempr Portal answers each one: the Dashboard, the log explorer, the request page, metadata, alerts and the weekly report.
- [When a provider retires an API](https://temprhq.io/blog/when-a-provider-retires-an-api): Perplexity replaced its Sonar chat completions with a new Agent API, retired on September 27. What changed for Tempr users, what we kept working, and how Tempr handles a provider retiring something more generally.

## Install

- [Visual Studio extension](https://marketplace.visualstudio.com/items?itemName=Tempr.TemprChat): Tempr Chat for Visual Studio 2022 and 2026, on the Visual Studio Marketplace
- [VS Code extension](https://marketplace.visualstudio.com/items?itemName=Tempr.tempr-chat): Tempr Chat for VS Code, on the Visual Studio Marketplace
- [VS Code extension (Open VSX)](https://open-vsx.org/extension/tempr/tempr-chat): The same VS Code extension, on the Open VSX Registry
- [JetBrains plugin](https://plugins.jetbrains.com/plugin/33782-tempr-chat): Tempr Chat for IntelliJ IDEA, PyCharm, WebStorm, Rider, and the other JetBrains IDEs
- [Tempr CLI (install script)](https://temprhq.io/cli#install): Self-contained builds for Windows, macOS and Linux: curl -fsSL https://temprhq.io/cli/install.sh | sh, or irm https://temprhq.io/cli/install.ps1 | iex
- [Tempr CLI (npm)](https://www.npmjs.com/package/@tempr/cli): npm i -g @tempr/cli
- [Tempr CLI (Homebrew)](https://github.com/chrissalomon/homebrew-tempr): Homebrew tap for macOS and Linux: brew tap chrissalomon/tempr, brew trust chrissalomon/tempr, brew install tempr
- [Tempr CLI (Scoop)](https://github.com/chrissalomon/scoop-tempr): Scoop bucket for Windows: scoop bucket add tempr https://github.com/chrissalomon/scoop-tempr, then scoop install tempr
- [Tempr CLI (NuGet)](https://www.nuget.org/packages/Tempr.Cli): .NET global tool: dotnet tool install -g Tempr.Cli (needs the .NET SDK)

## Docs: Getting Started

- [Introduction](https://temprhq.io/docs/): Tempr documentation: quickstarts, guides, and API reference for the Chat IDE extension, the Tempr CLI, and the Tempr Gateway API.
- [Quickstart: Chat in your IDE](https://temprhq.io/docs/quickstart-ide): Install the Tempr extension in Visual Studio, VS Code, or a JetBrains IDE, activate a license, add a provider key, and send your first chat message.
- [Quickstart: Gateway API](https://temprhq.io/docs/quickstart-gateway): Get a Tempr Gateway virtual key and make your first OpenAI-compatible request in under 5 minutes.
- [Quickstart: Tempr CLI](https://temprhq.io/docs/quickstart-cli): Install the Tempr CLI, sign in with your license key, and run your first agent turn from the terminal.
- [Teams & SSO](https://temprhq.io/docs/teams-sso): Tempr for organizations - centralized billing, seat management, shared personas, SSO (SAML/OIDC), and org-managed virtual keys with per-member limits.
- [Org rules & skills](https://temprhq.io/docs/org-rules-and-skills): Publish your organization's rules and skills once, and every Tempr agent follows them: required and recommended rules, SKILL.md skills, personas and slash commands, by group.

## Docs: Chat (IDE)

- [Providers & BYOK](https://temprhq.io/docs/providers): The full list of AI providers Tempr supports, how bring-your-own-key works, and how provider keys are stored and encrypted.
- [Custom models](https://temprhq.io/docs/custom-models): Run models on your own OpenAI-compatible server (vLLM, SGLang, LiteLLM) through Tempr Gateway, Chat, and the CLI: endpoints, keys, model names, pricing, and server notes.
- [Agent modes & personas](https://temprhq.io/docs/agent-modes): Interactive, Bypass, and Autopilot agent modes; named agent personas; and multi-agent delegation in Tempr Chat.
- [Inline completion](https://temprhq.io/docs/ide-completion): Ghost-text inline completion in Visual Studio, VS Code, and JetBrains IDEs - same BYOK providers as chat, per-scenario models, included on every plan.
- [Sessions, history & cloud sync](https://temprhq.io/docs/ide-sessions): How Tempr saves chat sessions, what syncs across machines on Standard and above, and why agent runs survive dropped connections.
- [MCP servers](https://temprhq.io/docs/mcp): Connect Model Context Protocol (MCP) servers to extend the Tempr agent with your own tools and data sources.
- [MCP Connectors for Claude, ChatGPT and More](https://temprhq.io/docs/mcp-connectors): Add your MCP servers to claude.ai, Claude Desktop, ChatGPT or Claude Code as one connector URL. People sign in to Tempr, and every tool call goes through your credentials, Tool Packs, guardrails and logs.

## Docs: CLI

- [CLI Command Reference](https://temprhq.io/docs/cli-reference): Every Tempr CLI command, flag, and REPL slash command - one-shot prompts, the interactive REPL, piped headless mode, session resume, and the agent tool set.
- [CLI: Config & troubleshooting](https://temprhq.io/docs/cli-config): Signing in and out, picking a default model, and diagnosing the failures that actually happen - with the fixes.
- [CLI Scripts, Pipes & CI](https://temprhq.io/docs/cli-scripts): Use the Tempr CLI headless: pipe prompts in, read output out, stream JSON events, and run it in scripts and CI with per-run model and auto-approve flags.
- [Tempr in Your Editor (ACP)](https://temprhq.io/docs/cli-editors): Run Tempr as an Agent Client Protocol agent in Zed, JetBrains AI Assistant, Neovim and Emacs: tool calls with diffs, approvals, sessions and settings in the editor.

## Docs: Gateway API

- [Authentication & virtual keys](https://temprhq.io/docs/gateway-auth): How Tempr Gateway authentication works: virtual keys, model allowlists, budgets, and rate limits.
- [Use Claude, ChatGPT and Coding Plans Through the Gateway](https://temprhq.io/docs/ai-plans): Run Claude Code on your Claude plan, Codex on your ChatGPT plan, and 30 coding plans such as Z.AI, Kimi and MiniMax through Tempr Gateway. The plan pays; you keep request logs, key limits, guardrails and a fallback for when it runs out.
- [Gateway: Virtual keys cookbook](https://temprhq.io/docs/gateway-cookbook): Practical virtual key recipes for the Tempr Gateway - one key per environment, self-capping budgets, model allowlists, instant revocation.
- [Chat Completions](https://temprhq.io/docs/gateway-chat-completions): Full reference for POST /v1/chat/completions on Tempr Gateway: parameters, streaming, tool calling, response shape, and headers.
- [Embeddings](https://temprhq.io/docs/gateway-embeddings): Full reference for POST /v1/embeddings on Tempr Gateway: parameters, providers and models, dimensions, response shape, and what applies to embeddings.
- [Models, limits & logs](https://temprhq.io/docs/gateway-reference): The GET /v1/models endpoint, rate limits, monthly quotas, metered overage, fallback chains, prompt and response caching, request logs, content capture, and Enterprise log streaming on Tempr Gateway.
- [Errors](https://temprhq.io/docs/gateway-errors): Tempr Gateway error shape and status codes: auth, rate limits, quota, model allowlist, and upstream provider errors.
- [Management API](https://temprhq.io/docs/gateway-management-api): Create, change and revoke Tempr Gateway virtual keys and read usage from your own scripts, with a management key.

## Docs: Guides

- [Claude Code Integration](https://temprhq.io/docs/guide-claude-code): Point Claude Code at Tempr Gateway and run GPT-5, Gemini, or any model on your allowlist - not just Claude - via ANTHROPIC_BASE_URL and ANTHROPIC_AUTH_TOKEN.
- [Claude Desktop Integration](https://temprhq.io/docs/guide-claude-desktop): Run Claude Desktop through Tempr Gateway with its third-party inference mode, and roll the setup out to every Mac and Windows PC with a device-management profile.
- [Cursor Integration](https://temprhq.io/docs/guide-cursor): Override Cursor's OpenAI base URL to run any model on your own gateway allowlist, billed through your own BYOK keys.
- [Cline Integration](https://temprhq.io/docs/guide-cline): Point Cline at Tempr Gateway to reach every model you've configured through one provider entry and one key.
- [Roo Code Integration](https://temprhq.io/docs/guide-roo-code): Point Roo Code at Tempr Gateway to reach every model on your allowlist through one OpenAI Compatible provider entry.
- [Kilo Code Integration](https://temprhq.io/docs/guide-kilo-code): Point Kilo Code at Tempr Gateway to reach every model on your allowlist through one custom provider entry.
- [Continue Integration](https://temprhq.io/docs/guide-continue): Point Continue at Tempr Gateway via config.yaml's apiBase field to reach every model on your allowlist with one key.
- [Zed Integration](https://temprhq.io/docs/guide-zed): Add Tempr Gateway as a custom provider in Zed to list every model on your allowlist under one key.
- [Codex CLI Integration](https://temprhq.io/docs/guide-codex-cli): Point Codex CLI at Tempr Gateway as a custom model provider and run Claude, Gemini, DeepSeek, or OpenAI's own codex models over the Responses API.
- [Aider Integration](https://temprhq.io/docs/guide-aider): Point Aider at Tempr Gateway via OPENAI_API_BASE to reach every model on your allowlist under one key.
- [Crush Integration](https://temprhq.io/docs/guide-crush): Add Tempr as a custom provider in Crush to reach every model on your allowlist under one config block.
- [OpenCode Integration](https://temprhq.io/docs/guide-opencode): Add Tempr as a custom provider in OpenCode to reach every model on your allowlist under one config block.
- [n8n Integration](https://temprhq.io/docs/guide-n8n): Route n8n's OpenAI node through Tempr Gateway to call Claude, Gemini, or any model in your workflows - not just OpenAI's.
- [GitHub Copilot App Integration](https://temprhq.io/docs/guide-github-copilot): Use Tempr Gateway as a BYOK provider so GitHub Copilot's agent sessions can run any model, not just Copilot's curated list.

## Optional

- [Contact](https://temprhq.io/contact): Contact the Tempr team - questions about pricing, a Teams plan, Gateway plans, or something not working as expected. We reply to every message.
- [Terms of Service](https://temprhq.io/terms): Terms of Service for Tempr, the bring-your-own-key platform for AI coding agents and model access: Tempr Chat, CLI, Code, and Gateway.
- [Privacy Policy](https://temprhq.io/privacy): Privacy Policy for Tempr, a bring-your-own-key AI coding assistant for Visual Studio, VS Code, and JetBrains IDEs.
- [Cookie Policy](https://temprhq.io/cookies): Cookie Policy for the Tempr marketing website and customer portal.
- [EULA](https://temprhq.io/eula): End User License Agreement for Tempr software - the extensions for Visual Studio, VS Code, and JetBrains IDEs, the Tempr CLI, and related clients.