On June 10, 2026, GitHub confirmed that Copilot would move to usage-based billing starting June 15, 2026. The announcement appeared on the official pricing page and in a changelog sent to account owners. The old flat-rate promise disappeared. Copilot Pro no longer meant unlimited completions and chat messages. Instead, GitHub attached a meter to premium requests. Pro subscribers received a new $10 monthly base price and 500 included premium requests. Additional premium requests cost $0.04 each. Business and Enterprise plans received separate request pools with per-seat overage fees. The company described the change as a way to align price with compute cost. Many developers described it as a price increase hiding behind a lower entry fee.

Free users were not spared. Copilot Free kept a daily chat limit of 50 messages and a small completion allowance. Premium model access stayed gated behind paid plans. The change hit individual developers, small teams, and enterprise accounts differently. Copilot Business moved to $19 per user per month with 1,000 premium requests included. Copilot Enterprise moved to $39 per user per month with 2,000 requests. Users who burned through their pool paid overage at the same $0.04 per premium request. The pricing shift followed a wave of coding tool changes in May and June 2026. OpenAI expanded the Codex free tier for agentic coding, and Anthropic replaced flat-rate Claude access with a credit pool on June 15, 2026. GitHub chose a different path: a low base price with a meter.

The competitive context mattered. AI coding assistants had been racing to offer unlimited or near-unlimited plans. Anthropic ended its agent subsidy on June 15. Google shut down Gemini Code Assist for free users earlier in June. Cursor and Windsurf tightened free tiers. GitHub Copilot entered that environment with a usage-based plan that looked cheaper up front but penalized heavy users through overage and multipliers. The announcement raised immediate questions about hidden costs. A premium request could include multiple model calls under the hood. GitHub said agentic coding tasks would consume more premium requests per turn. That multiplier became the center of developer anger within hours. The story was not just a price change. It was a signal that unlimited AI coding access was ending across the industry. The AI API free tiers and limits report documented similar tightening across major providers.

Developers who relied on Copilot for daily work faced a new math problem. A heavy user generating 2,000 premium requests in a month paid $10 plus 1,500 overage requests. That total reached $70 per month before taxes. The previous $19 Pro plan included flat access without per-request fees, though some limits applied under fair use. Business accounts with multiple seats saw unpredictable costs. Teams had to estimate request volume or risk surprise bills. The major AI coding tools pricing overhaul showed that GitHub was not alone. Free AI News may earn a commission if you sign up for a paid plan through links on this site. This does not affect our reporting.

How Do the Top Options Compare?

Plan or Tool Best For Monthly Base Included Premium Requests Overage Cost Key Limit
GitHub Copilot Free Occasional suggestions $0 50 chat messages per day N/A No premium models
GitHub Copilot Pro Individual devs $10 500 $0.04 per request Agentic multiplier applies
GitHub Copilot Business Small teams $19 per user 1,000 per user $0.04 per request Per-seat overage
Cursor AI coding alternative $20 500 fast requests $0.04 per fast request Unlimited slow requests

Prices verified from GitHub’s public pricing page on June 10, 2026. Overage costs vary by model and task. Free tier limits reset daily.

1. GitHub Copilot Free , Best for occasional AI code suggestions

GitHub kept a free tier after June 15, 2026, but it became more restrictive. Free users received 50 chat messages per day and a limited number of basic code completions. Premium model access required a paid plan. The free tier existed as an entry point, not a daily driver. Many free users noticed the limits compared to alternatives. Cursor and Windsurf also tightened their free tiers in June 2026, as covered in cursor-windsurf-zed-free-tier-2026.

GitHub Copilot Free did not include usage-based billing because there was no usage to bill. Instead, the limits reset daily. Developers who needed more than light autocomplete hit the cap quickly. The free tier still supported basic Python, JavaScript, TypeScript, and a handful of other languages. It did not include Copilot agent mode or premium reasoning. GitHub marketed the free tier as a way to try Copilot before paying.

For casual users, the free tier remained useful. For working developers, it became a preview. The change reflected the broader shift documented in AI API free tiers and limits 2026. Free access was not gone, but it was shrinking.

Key strengths:

  • ✅ No monthly cost
  • ✅ 50 chat messages per day
  • ✅ Basic completions in popular languages
  • ✅ No credit card required
  • ❌ No premium model access
  • ❌ Daily cap blocks full workdays
  • ❌ No agentic coding

Who it’s for: Choose GitHub Copilot Free if you only need light autocomplete and want to test Copilot before paying.

2. GitHub Copilot Pro , Best for individual developers who want premium models

The June 15, 2026 change transformed Copilot Pro from a flat $19 per month plan into a $10 base plan with 500 premium requests included. Additional premium requests cost $0.04 each. A month with 1,000 premium requests cost $30. A month with 2,000 cost $70. GitHub published the new terms on its pricing page and changelog. The company said the change aligned cost with compute. Developers saw it differently.

Agentic coding tasks carried a multiplier. GitHub confirmed that a single agentic turn could consume multiple premium requests. The exact multiplier varied by model and task. This detail triggered the backlash covered in GitHub Copilot multiplier backlash. A developer using Copilot agent mode for a complex refactor could burn through ten or twenty requests in one session. The meter made costs hard to predict.

The lower entry price looked attractive. But heavy users paid far more than before. The old unlimited plan had been a key selling point. GitHub ended that promise. The change placed Copilot Pro in the middle of the AI coding tools pricing overhaul. Individual developers had to monitor usage or face higher bills. GitHub acknowledged the concern but did not reverse the decision.

Key strengths:

  • ✅ Lower $10 monthly base
  • ✅ 500 included premium requests
  • ✅ Access to premium models
  • ✅ Pay only for what you use
  • ❌ Overage adds up quickly
  • ❌ Agentic multiplier raises costs
  • ❌ No unlimited plan

Who it’s for: Choose Copilot Pro if you use fewer than 500 premium requests per month and want premium model access at a lower base price.

3. GitHub Copilot Business , Best for teams that need admin controls

Copilot Business moved to $19 per user per month on June 15, 2026. Each user received 1,000 premium requests included. Additional requests cost $0.04 each. Organizations gained admin controls, license management, and usage reporting. The per-seat price stayed competitive, but the hidden overage risk grew. A team of ten developers using 2,000 premium requests each per month paid $190 base plus $400 in overage fees. That monthly total reached $590 before taxes.

The backlash from business accounts was documented in developer outcry over GitHub Copilot hidden costs. Finance teams were not prepared for variable billing. GitHub offered usage dashboards and alerts, but the damage was done. The old Business plan included flat access with fewer surprises. The new plan forced teams to forecast request volume.

Despite the change, Copilot Business retained strong integration with GitHub repositories, pull requests, and CI workflows. For teams already embedded in GitHub, switching away carried real costs. But the pricing shift made Cursor and Windsurf alternatives more attractive for cost-sensitive teams.

Key strengths:

  • ✅ 1,000 premium requests per user
  • ✅ Admin controls and usage reports
  • ✅ Tight GitHub workflow integration
  • ✅ Per-seat pricing predictable at low usage
  • ❌ Overage costs across teams add up
  • ❌ No unlimited team plan
  • ❌ Usage forecasting required

Who it’s for: Choose Copilot Business if your team lives in GitHub and you can monitor request volume to avoid surprise overage bills.

4. Cursor , Best alternative with predictable AI coding pricing

Cursor entered June 2026 as the main alternative for developers unhappy with GitHub Copilot. Its $20 per month Pro plan included 500 fast requests. Additional fast requests cost $0.04 each. Slow requests remained unlimited. That structure was similar to Copilot Pro, but Cursor offered a more generous unlimited slow pool and a polished agent mode. The comparison was not simple, but many developers considered it a better value.

The major AI coding tools pricing overhaul showed Cursor competing directly with GitHub and OpenAI. OpenAI’s Codex free tier for agentic coding, covered in ChatGPT Codex free tier agentic coding 2026, added more pressure. Cursor did not require GitHub integration, which attracted developers using GitLab or self-hosted repositories.

Cursor’s main weakness was ecosystem lock-in. It was a standalone editor, not a plugin inside an existing IDE. Teams standardized on Visual Studio Code found Copilot easier to adopt. But developers who wanted transparent pricing and strong agent features increasingly chose Cursor. The June 2026 changes to Copilot pushed more users to compare the two tools side by side.

Key strengths:

  • ✅ Unlimited slow requests on Pro
  • ✅ Strong agent mode
  • ✅ Works without GitHub
  • ✅ Clear fast request pricing
  • ❌ Fast requests can run out
  • ❌ Standalone editor requires migration
  • ❌ Team pricing less mature

Who it’s for: Choose Cursor if you want a Copilot alternative with unlimited slow requests and do not need deep GitHub integration.

Frequently Asked Questions

When did GitHub Copilot usage-based billing start?

GitHub Copilot usage-based billing began on June 15, 2026, after GitHub announced the change on June 10, 2026 via its pricing page and changelog.

What did Copilot Pro cost before June 2026?

Copilot Pro previously cost $19 per month with flat access to chat and completions under fair use. After June 15, 2026, it became a $10 base plan with 500 premium requests and $0.04 per extra request.

What counted as a premium request?

GitHub counted requests that use premium models or agentic features. A single agentic turn could consume multiple premium requests depending on model and task.

Did the free tier change?

Yes. Copilot Free kept a daily limit of 50 chat messages and limited basic completions. Premium model access remained paid only.

Are there cheaper alternatives?

Yes. Cursor offered a $20 plan with unlimited slow requests, and OpenAI expanded a Codex free tier for agentic coding in June 2026.

What was the developer backlash?

Developers objected to unpredictable overage costs and hidden multipliers. GitHub added usage dashboards but did not restore unlimited flat pricing.

What Should You Remember?

  • Metered billing: GitHub Copilot Pro became a $10 base plan with 500 included premium requests starting June 15, 2026.
  • Overage cost: Each additional premium request cost $0.04, so 1,500 extra requests added $60 per month.
  • Agentic multiplier: A single agentic task could burn multiple premium requests, making costs hard to predict.
  • Free tier limits: Free users kept 50 chat messages per day and lost access to premium models.
  • Team risk: Business accounts faced variable per-seat overage, which angered finance teams.
  • Alternatives: Cursor and OpenAI Codex offered different pricing structures for developers fleeing unlimited-plan removal.

Free AI News is an independent editorial publication. Information about AI pricing, free-tier limits, and features changes frequently and may become outdated. Always verify current details through the vendor’s official pages. Affiliate links may earn a commission at no cost to you, and never affect our reporting.