Compare · Checked

Tempr vs GitHub Copilot

Both run in Visual Studio, VS Code, and JetBrains IDEs, and both now accept your own API key. The difference is how much your key covers: in Tempr it powers chat, the agent, and inline completion; in Copilot it covers chat and the agent, while code completions always run on Copilot's own models.

The short answer

Side by side

TemprGitHub Copilot
IDEsVisual Studio 2022 and 2026, VS Code, JetBrains IDEsVS Code, Visual Studio, JetBrains IDEs, and other editors
Your own key for chat and agentYes, on every planYes in VS Code, with no GitHub account needed; in preview in Visual Studio (new Agent only) and JetBrains IDEs
Your own key for inline completionYes, in all three IDEsNo; completions run on Copilot's models and need a GitHub account
Providers on your own keyAbout 30: first-party providers, cloud platforms, inference hosts, and OpenRouter; plus your own OpenAI-compatible endpoints on ProIn VS Code: Anthropic, Azure, Google Gemini, OpenAI, Ollama, and custom endpoints
Built-in modelsNone; every model runs on your keyOpenAI, Anthropic, Google, and other models, billed in AI Credits
Price for one personStandard $15/mo, Pro $20/mo; 7-day free trialFree; Pro $10/mo, Pro+ $39/mo, Max $100/mo
TeamsFrom $35/seat/mo (10-seat minimum), with SSOBusiness $19/seat, Enterprise $39/seat
How usage is paid forFlat subscription; you pay your providers directly at their own rates, with no markupChat, the CLI, and the cloud agent draw on metered AI Credits; completions don't. On your own key, your provider bills you
AgentAgent in the IDE with Interactive, Bypass, and Autopilot modes, personas, delegation, and MCP servers; the same agent in the terminal (CLI included)Agent mode in the IDE, MCP, a cloud agent that runs in GitHub Actions, and Copilot CLI
Where requests go on your own keyThrough Tempr's servers to your providerChat goes to your provider directly; completions, semantic search, and built-in models go through GitHub
Training on your dataYour prompts, code, and chats aren't used to train modelsOn Free, Pro, and Pro+, interaction data trains GitHub's models unless you opt out

What your own key covers

Copilot added bring-your-own-key to VS Code, where it works without signing in to GitHub, and has it in preview in Visual Studio and JetBrains IDEs. In VS Code it applies to the chat experience and utility tasks; inline suggestions (code completions), semantic search, and features that rely on embeddings still need a GitHub account and run on GitHub's service. In Visual Studio it works only in the new Agent (Preview).

In Tempr there are no built-in models at all, so your key is behind everything: chat, the agent, and inline completion, in Visual Studio, VS Code, and JetBrains IDEs alike. You can set a fast, inexpensive model for completion and a stronger one for chat and the agent, all on the same key.

Pricing

Copilot's plans include an allowance of AI Credits for its built-in models, and since June 2026 chat, the CLI, and the cloud agent draw on those credits, with extra usage billed on top; code completions and next edit suggestions don't use credits. Tempr is a flat subscription — Standard $15 or Pro $20 a month — and every request goes out on your key at your provider's own price, with no markup. If you already pay for provider API keys, Tempr puts them to work in all three IDEs; if you don't, Copilot's bundled models are simpler to start with.

Where your code goes

With Copilot's bring-your-own-key chat, requests go from your machine to your provider; completions, semantic search, and the built-in models go through GitHub. On the Free, Pro, and Pro+ plans, GitHub uses interaction data to train its models unless you opt out.

Tempr relays every request through its servers to your provider. The agent's tool-calling loop runs server-side so a turn survives a dropped connection, and each run is deleted 72 hours after it finishes. Tempr doesn't use your prompts, code, or chats to train models. The Privacy Policy has the details.

Already on Copilot?

You don't have to choose. The Tempr Gateway can be Copilot's bring-your-own-key provider, so Copilot's agent sessions can use any model on your Gateway allowlist.

Questions

Yes, for chat and the agent. In VS Code it works without a GitHub account; in Visual Studio (in the new Agent) and JetBrains IDEs it's in preview. Code completions still run on Copilot's own models.

No. Copilot's code completions run on its own models and need a GitHub account, even when chat uses your key. In Tempr, inline completion runs on your own key in Visual Studio, VS Code, and JetBrains IDEs.

It depends on how you work. Copilot Pro is $10 a month and Tempr Standard $15, but with Tempr you also pay your provider for the tokens you use, at the provider's own rates, while Copilot's price includes an allowance for its built-in models and charges for heavier use. If you already pay for API keys, Tempr puts them to work across all three IDEs.

Yes. Tempr is a separate extension, and the Tempr Gateway can act as Copilot's bring-your-own-key provider, so Copilot's agent can use any model on your Gateway allowlist. See the GitHub Copilot guide.

How we checked

Everything about GitHub Copilot comes from its own pricing pages, documentation, and release notes, checked on September 21, 2026. These products change quickly, so prices and features may have moved since. Spotted something out of date? Tell us and we'll fix it.

Sources: Copilot plans · language models in VS Code · bring your own key · Visual Studio 2026 release notes · JetBrains BYOK · usage-based billing · cloud agent · interaction data policy

Try Tempr on your own keys

Every paid Chat plan starts with a 7-day free trial, and the CLI is included.