GitHub Copilot is moving to usage-based billing
GitHub Copilot is shifting to usage-based billing, a move GitHub frames as aligning pricing with the compute demands of its evolving agentic capabilities. While base subscription prices remain unchanged, the new credit system and dramatically increased model multipliers have led many users to perceive this as a significant, hidden price hike. The developer community is now grappling with what this means for their workflows and budgets, openly questioning Copilot's value compared to direct API access or alternative AI coding tools.
The Lowdown
GitHub Copilot is transitioning to a usage-based billing model, effective June 1, 2026. This change replaces the previous premium request unit (PRU) system with GitHub AI Credits, which are consumed based on token usage. GitHub states this shift is necessary due to the increased compute and inference demands of Copilot's advanced 'agentic' features and to ensure the service's long-term sustainability.
Key aspects of the change include:
- Base subscription prices for all plans (Pro, Pro+, Business, Enterprise) will not change numerically.
- Each plan will now include a monthly allotment of AI Credits equivalent to the subscription cost (e.g., $10 plan gets $10 in credits).
- Usage will be calculated based on tokens, with significant model multiplier increases (e.g., Opus 3x to 27x, GPT 5.4 1x to 6x) for advanced models.
- Code completions and Next Edit suggestions remain included and do not consume AI Credits.
- Fallback experiences for exhausted PRUs are being removed, replaced by hard credit limits or admin budget controls.
- Copilot code review will additionally consume GitHub Actions minutes.
- Organizations will benefit from pooled included usage and new budget controls to manage spend.
GitHub assures users of full control over spending with visibility tools and the option to purchase additional credits. However, the community interprets these changes, particularly the steep increase in model multipliers, as a substantial effective price hike, pushing many to reconsider the value proposition of their Copilot subscriptions.
The Gossip
Multiplier Mayhem
The most prominent discussion revolves around the drastic increase in model multipliers, which effectively make advanced Copilot features significantly more expensive. Users highlight examples like Opus going from 3x to 27x, and GPT models seeing a 6x to 18x increase. This leads many to conclude that despite GitHub's assertion of unchanged plan prices, the cost-per-token for actual usage has skyrocketed, undermining the previous value proposition.
The Subsidy Shift
Many commenters acknowledge that the 'era of subsidized inference' for AI coding assistants is drawing to a close, recognizing that GitHub can no longer absorb the escalating costs of powerful AI models. While understanding the necessity of passing on these costs, there's a strong sentiment that the current implementation feels like a Bait and Switch, particularly given the perceived lack of competitive pricing compared to direct API access or other providers, even with GitHub's scale.
Considering Competitors
A significant portion of the discussion centers on users actively exploring or planning to switch to alternative AI coding tools and direct API access providers like OpenRouter or Codex CLI. The steep effective price increase and the feeling that Copilot no longer offers sufficient value for its premium features are major drivers. Users are questioning if Copilot's deep IDE integration alone justifies the new, higher effective cost per token.
Autocomplete's Anchor
Several developers suggest that with the new pricing structure, Copilot's primary value may now reside solely in its basic code completion and IntelliSense-like features. They indicate that for more complex 'agentic' coding sessions, the new token-based billing and increased multipliers make it economically unfeasible, pushing them to use other tools or direct model access for those advanced tasks.