Key Terms
- PRU (Premium Request Unit) - GitHub's current billing unit for Copilot. Each chat, agent mode turn, code review, or cloud agent task consumes one or more PRUs depending on the model and feature. A PRU multiplier applies per model (e.g., Opus 4.7 = 7.5x PRU). Source: GitHub – Copilot Requests
- GitHub AI Credits - Replacing PRUs on June 1, 2026. Token-based billing where usage is charged based on actual input/output/cached tokens at per-model API rates. Source: GitHub – Github Copilot Is Moving To Usage Based Billing
- Model multiplier - a factor applied to each premium request based on the model used. A 7.5x multiplier means one request to that model counts as 7.5 PRUs against the monthly allowance. Source: GitHub – Models And Pricing
- IP indemnity - the provider covers legal costs if their AI output infringes a third party's copyright. Included in Business and Enterprise plans. Source: Blogs – Copilot Copyright Commitment Ai Legal Concerns
- SAML SSO - employees log in via their corporate identity provider (Okta, Azure AD) instead of separate passwords. Source: GitHub – Identity And Access Management
- Agent mode - Copilot operates autonomously in the IDE, reading files, running commands, and making multi-step edits. Source: GitHub – Copilot Agent Mode
- MCP servers (Model Context Protocol) - a standard for connecting AI agents to external tools and data sources. Copilot integrates with MCP servers in agent mode. Source: GitHub – Using Mcp Servers With Copilot
Latest Changes
First report for this supplier. All models, plans, and pricing are listed as current state.
- Plan change: Claude Opus models (4.5, 4.6, 4.7) removed from Copilot Pro ($10/mo) on April 20. Pro+ ($39/mo) now required for Opus access. Source: GitHub – 2026 04 20 Changes To Github Copilot Plans For Individuals
- Plan change: Self-serve Copilot Business signups paused April 22. Source: GitHub – 2026 04 22 Pausing New Self Serve Signups For Github Copilot Business
- Plan change: New Copilot Pro trials paused April 10. Opus 4.6 Fast mode retired from Pro+. Source: GitHub – 2026 04 10 Pausing New Github Copilot Pro Trials
- Price change (upcoming): All Copilot plans transitioning from PRU-based to token-based billing (GitHub AI Credits) on June 1, 2026. Base plan prices unchanged but actual costs will increase for frontier model users. Source: GitHub – Github Copilot Is Moving To Usage Based Billing
- New model: Claude Opus 4.7 available April 16 at 7.5x PRU multiplier (described as "introductory" rate). Source: GitHub – 2026 04 16 Claude Opus 4 7 Is Generally Available
- Feature added: Copilot CLI now supports BYOK and local models (April 7). Source: GitHub – 2026 04 07 Copilot Cli Now Supports Byok And Local Models
- Deprecation (upcoming): Copilot code review will consume GitHub Actions minutes starting June 1. Source: GitHub – 2026 04 27 Github Copilot Code Review Will Start Consuming Github Actions Minutes On June 1 2026
- Feature removed: GPT-5.3-Codex removed from Student plan model picker (April 27). Source: GitHub – 2026 04 27 Copilot Student Gpt 5 3 Codex Removal From Model Picker
Plans
Individual Plans
| Plan | Price | Premium Requests / AI Credits | Key Inclusions |
|---|---|---|---|
| Free | $0/mo | 50 premium requests/mo, 2,000 completions/mo | Agent mode (50/mo), chat, CLI, models: Haiku 4.5, GPT-5 mini + more |
| Pro | $10/mo | 300 premium requests/mo | Unlimited GPT-5 mini agent/chats, unlimited inline suggestions, cloud agent, code review, Claude and Codex access, 20 models available |
| Pro+ | $39/mo | 1,500 premium requests/mo | All models including Opus 4.7, GitHub Spark (preview), delegate to third-party agents (Claude, Codex) |
Source: GitHub – Plans
Important: Pro and Pro+ are temporarily unavailable for new signups as of April 2026 (reliability measures ahead of billing transition). Source: GitHub – Changes To Github Copilot Individual Plans
Business Plans
| Plan | Price | Premium Requests / AI Credits | Key Inclusions |
|---|---|---|---|
| Business | $19/user/mo | Included (undisclosed PRU count) | Everything in Pro + IP indemnity, SAML SSO, user management, policies, usage metrics |
| Enterprise | $39/user/mo | Included (undisclosed PRU count) | Everything in Business + codebase indexing, custom models, GitHub.com chat, knowledge base, third-party agent delegation |
Important: Self-serve Business signups paused as of April 22, 2026. Source: GitHub – 2026 04 22 Pausing New Self Serve Signups For Github Copilot Business
Additional Premium Requests
$0.04 per additional premium request (current model). Under token-based billing (June 1), additional usage charged at published API rates.
Available Models (All Plans)
| Provider | Models |
|---|---|
| Anthropic | Haiku 4.5, Sonnet 4, Sonnet 4.5, Sonnet 4.6, Opus 4.5, Opus 4.6, Opus 4.6 (fast mode) (Preview), Opus 4.7 |
| Gemini 2.5 Pro, Gemini 3 Flash (Preview), Gemini 3.1 Pro (Preview) | |
| OpenAI | GPT-5 mini, GPT-5.2, GPT-5.2-Codex, GPT-5.3-Codex, GPT-5.4, GPT-5.4 mini, GPT-5.5 |
| xAI | Grok Code Fast 1 |
| GitHub | Raptor mini (Preview) |
Note: Opus models removed from Pro plan as of April 20, 2026. Pro+ ($39/mo) now required for Opus access. Source: GitHub – 2026 04 20 Changes To Github Copilot Plans For Individuals
API Pricing
GitHub Copilot does not offer a standalone API. Usage is billed through the subscription plans described above. Under the upcoming token-based billing model (June 1, 2026), usage will be charged at per-model API rates via GitHub AI Credits.
Model Performance / Benchmarks
GitHub does not publish independent benchmark scores for Copilot as a product. Copilot uses models from multiple providers; benchmark scores for those models are published by their respective providers. See the Anthropic, OpenAI, Google, and xAI raw reports for model-level benchmarks.
GitHub reported that Gemini 3 Pro delivers 35% higher accuracy vs Gemini 2.5 Pro in VS Code testing. Source: Deepmind – Gemini
Latest News
Token-Based Billing Announcement (April 27, 2026)
Major announcement: all Copilot plans transitioning to token-based billing on June 1, 2026, replacing premium request units with GitHub AI Credits. Base plan prices unchanged. See Token-Based Billing section above for details.
Source: GitHub – Github Copilot Is Moving To Usage Based Billing
Opus Removed from Pro Plan (April 20, 2026)
Claude Opus models (4.5, 4.6, 4.7) removed from Copilot Pro ($10/mo). Pro+ ($39/mo) now required for Opus access. Opus 4.7 set at 7.5x PRU multiplier, described as an "introductory" rate.
Source: GitHub – 2026 04 20 Changes To Github Copilot Plans For Individuals
Self-Serve Business Signups Paused (April 22, 2026)
New self-serve Copilot Business plan purchases temporarily suspended.
Source: GitHub – 2026 04 22 Pausing New Self Serve Signups For Github Copilot Business
Pro Trials Paused, Opus 4.6 Fast Retired (April 10, 2026)
New Copilot Pro trials paused. Opus 4.6 Fast mode retired from Pro+.
Source: GitHub – 2026 04 10 Pausing New Github Copilot Pro Trials
Opus 4.7 Launch (April 16, 2026)
Claude Opus 4.7 available on Copilot at 7.5x multiplier.
Source: GitHub – 2026 04 16 Claude Opus 4 7 Is Generally Available
Copilot CLI BYOK and Local Models (April 7, 2026)
Copilot CLI now supports Bring Your Own Key (BYOK) and local models.
Source: GitHub – 2026 04 07 Copilot Cli Now Supports Byok And Local Models
Code Review Consuming Actions Minutes (April 27, 2026)
Starting June 1, Copilot code review will consume GitHub Actions minutes in addition to AI Credits.
Student GPT-5.3-Codex Removed (April 27, 2026)
GPT-5.3-Codex removed from Student plan model picker.
Source: GitHub – 2026 04 27 Copilot Student Gpt 5 3 Codex Removal From Model Picker
Community Signals
Token-Based Billing Reaction
HN announcement thread: 747 points, 549 comments. Dominant community sentiment: the subsidized inference era is ending.
- User calculated a single 3x multiplier Opus request consumed tokens equivalent to ~$260 at API prices, costing the user $0.12. Source: News – Item
- 600%+ price increase reported for many workflows. Source: News – Item
- Multiple users recommend switching to OpenRouter for direct API access at the same cost with no monthly expiration. Source: News – Item
- Community consensus: "This is the best AI programming will be. From here on the enshitification starts and the prices go up." Source: News – Item
Quality Degradation After Opus Removal
Auto-router quality degraded after Opus models were removed from Pro. "Today was my first time using copilot since the change... It always worked well. Today it sucks, because they took away the option for the autorouter to use the best models." GPT-5.3 (new default) perceived as downgrade despite being SOTA weeks ago. Haiku deemed insufficient for complex work. Source: News – Item
Enterprise Users Choosing Alternatives
Enterprise user: "At my company basically everyone and their mother are using Claude Code via Bedrock, despite us having company-wide Windsurf, Copilot and ChatGPT Enterprise accounts." HN user: "Copilot is still an interesting case-study and (probably) failed predictor... consistently the weakest mainstream AI coding offering. Inferior to Cursor or Windsurf at editor completions, inferior to Codex, Claude, OpenCode." Source: News – Item
Price Comparisons
HN: "DeepSeek API pricing 300% difference vs Anthropic/OpenAI." One company reported average dev hitting $2,000/month in Claude Code costs. Source: News – Item
Enterprise Readiness
| Feature | Available? | Details |
|---|---|---|
| SSO (SAML) | Yes | Via GitHub Enterprise Cloud. Business and Enterprise plans. Source: GitHub – About Saml For Enterprise Iam |
| SSO (OIDC) | Yes | Via GitHub Enterprise Cloud with Entra ID. Source: GitHub – Identity And Access Management |
| SCIM | Yes | With Enterprise Managed Users or org-level SCIM with SAML SSO. Source: GitHub – Identity And Access Management |
| Audit logs | Yes | 180-day retention, streamable to SIEM, tracks Copilot policy/license changes and agent activity. Source: GitHub – Review Audit Logs |
| IP indemnity | Yes | Microsoft Copilot Copyright Commitment. Business and Enterprise plans. Source: Blogs – Copilot Copyright Commitment Ai Legal Concerns |
| Data residency | Yes | Via GitHub Enterprise Cloud with data residency. US + EU regions. +10% model multiplier applies. Source: GitHub – Github Copilot With Data Residency |
| HIPAA | Yes | GitHub Enterprise Cloud offers BAA; Copilot data handling covered under enterprise terms. Source: Ghec |
| Air-gapped / on-prem | No | Copilot is not available on GitHub Enterprise Server (self-hosted). Source: GitHub – Plans |
| SLA | Yes | GitHub Enterprise Cloud provides a 99.9% monthly uptime SLA. Copilot-specific SLA is not published separately. Source: GitHub – Enterprise |
| Admin controls (RBAC) | Yes | Centralized policy management, model access controls, content exclusion, user/license management, premium request budgets, MCP server management, custom agent management. Source: GitHub – Manage Enterprise Policies |
Transparency Gaps
| Gap | Details | Severity |
|---|---|---|
| Model multipliers not clearly disclosed | The exact PRU multiplier for each model is not shown on the pricing page. Users must dig into docs. For token-based billing, the per-token rates that will be used are not yet published. | High |
| "Introductory" Opus 4.7 multiplier | The 7.5x multiplier for Opus 4.7 is described as "introductory," implying it may increase. No commitment on duration. | High |
| PRU counts for Business/Enterprise | The number of included premium requests for Business ($19/mo) and Enterprise ($39/mo) plans is not disclosed on the pricing page. | High |
| Promotional credit duration | Business gets $30/mo and Enterprise gets $70/mo in promotional credits for June-August only. Unclear what happens after. | Medium |
| Status page methodology change | "Partial Outages" now count only 30% of duration, "Degraded Performance" counts 0%, inflating uptime numbers. Source: News – Item | Medium |
| Annual plan multiplier increases | Annual subscribers face model multiplier increases on June 1 but the specific new multipliers are buried in docs. | Medium |
| Free Copilot for maintainers cut | Free Copilot Pro for open-source maintainers cut for small projects without formal announcement. | Medium |
| Code review Actions minutes cost | No published estimate of how many Actions minutes a typical code review consumes. | Low |