Copilot Copilot Bearer (OpenAI-compat /chat/completions)
Microsoft's coding assistant — Claude, GPT-5 and Gemini behind one subscription.
⚠️Heads up: GitHub paused new sign-ups for the Pro, Pro+ and Student tiers on 2026-04-20 — citing that agentic workloads consume far more compute than the original pricing assumed. Existing Pro/Pro+ subscribers keep their plan; new individual users can only pick Free, Business or Enterprise. Opus 4.x has also been removed from Pro — only Pro+ keeps it.
GitHub Copilot is Microsoft's AI coding assistant, embedded in VS Code, JetBrains, Neovim and the GitHub web UI. Under the hood it routes prompts across Claude (Sonnet and Opus), GPT-5, GPT-4.1 and Gemini 2.5/3 — choose the model, pay once, get them all. Pricing ranges from a free tier (with monthly limits) up to Enterprise, with the value proposition being 'one frontier model per IDE seat'.
Strengths
- Multiple frontier models on one seat (Claude Sonnet/Opus, GPT-5, Gemini)
- Reuses your IDE quota — no extra subscription if you already pay for it
- OpenAI-compatible endpoint once you hold the session token
When to use it
- You already pay for GitHub Copilot in your editor
- You want one key for several frontier model families
- Comparing the same prompt across Claude / GPT / Gemini cheaply
Subscription plans
| Plan | Price | Quota | Available |
|---|---|---|---|
| Free | $0/mo | 2,000 completions / 50 premium requests per month | yes |
| Pro | $10/mo | Higher limits, Sonnet/GPT-5 (no Opus) | closed / sold out |
| Pro+ | $39/mo | Includes Claude Opus 4.7 + premium models | closed / sold out |
| Business | $19/mo | Per seat — admin controls, audit logs | yes |
| Enterprise | $39/mo | Per seat (+ $21 GH Enterprise Cloud) | yes |
Notes: From June 1, 2026 Copilot moves to token-based billing: Business $19/seat with $30 pooled credits, Enterprise $39/seat with $70 pooled credits (June–August promo). Verify on github.com/features/copilot/plans.
Referral: GitHub does not run a public referral program for Copilot. Discounts are limited to verified students/teachers (free), open-source maintainers (free), and GitHub Enterprise volume pricing.
Models tested on Copilot
Speed numbers below are specific to Copilot's routing and hardware. The same model may appear on other providers' pages with different throughput.
| Model | Best tok/s | Avg tok/s | Runs | Success | Longest output (chars) |
|---|---|---|---|---|---|
| gpt-5-mini | 113.3 | 110.3 | 4 | 100% | 3,516 |
| gemini-3-flash-preview | 107.9 | 92.9 | 4 | 100% | 3,754 |
| claude-haiku-4.5 | 104.8 | 99.2 | 4 | 100% | 1,531 |
| gpt-4o | 93.1 | 91.1 | 4 | 100% | 3,424 |
| gpt-4.1 | 92.5 | 76.2 | 4 | 100% | 2,800 |
| gemini-2.5-pro | 75.3 | 71.6 | 4 | 100% | 5,637 |
| grok-code-fast-1 | 73.5 | 41.8 | 4 | 100% | 2,892 |
| gpt-5.2 | 68.9 | 56.5 | 4 | 100% | 2,874 |
| gemini-3.1-pro-preview | 62.4 | 50.1 | 4 | 100% | 3,570 |
| claude-sonnet-4.5 | 59.2 | 53.8 | 4 | 100% | 1,645 |
| claude-sonnet-4.6 | 52.2 | 50.8 | 4 | 100% | 1,053 |
| claude-sonnet-4 | 51.2 | 49.4 | 2 | 100% | 1,780 |