--- title: GitHub Copilot switches to token-based billing: what... description: GitHub retired its premium request model on June 1, replacing it with token-based 'AI Credits' billing. Plan prices remain unchanged, but usage is now metered... canonical: https://daily.dev/posts/github-copilot-switches-to-token-based-billing-what-changes-and-what-doesn-t-dqgznlmcl twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: GitHub Copilot switches to token-based billing: what changes and what doesn't | daily.dev og:description: GitHub retired its premium request model on June 1, replacing it with token-based 'AI Credits' billing. Plan prices remain unchanged, but usage is now metered... og:url: https://daily.dev/posts/github-copilot-switches-to-token-based-billing-what-changes-and-what-doesn-t-dqgznlmcl og:image: https://api.daily.dev/og/posts/DqgZnLMCL.png og:image:alt: GitHub Copilot switches to token-based billing: what changes and what doesn't og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # GitHub Copilot switches to token-based billing: what changes and what doesn't **[Collections](https://daily.dev/sources/collections)** · 4 min read · 3 upvotes · 2 comments ## Summary GitHub retired its premium request model on June 1, replacing it with token-based 'AI Credits' billing. Plan prices remain unchanged, but usage is now metered across chat, agent mode, inline assists, and other interactions — not code completions. Credit allotments range from $15/month (Pro) to $200/month (Copilot Max), with model costs spanning $0.25 to $5.00 per million tokens. Agentic workflows are the primary cost driver, consuming roughly 30x more tokens than regular chat. Real-world reports show costs jumping from $50 to $3,000/month for heavy agentic use. Enterprise admins currently lack granular model-permission controls, leaving cost management to manual caps and usage API monitoring. Alternatives like Claude Code ($20–$200/month) and BYOK multi-model setups are gaining traction among users seeking more predictable costs. ## Content ## What changed On June 1, GitHub retired Copilot's premium request model and replaced it with a token-based billing system using "AI Credits." Plan prices stayed the same, but usage is now metered by token consumption across input, output, and cached data. The credit allotments by plan: - **Pro** ($10/month): $15 in monthly credits - **Pro+** ($39/month): $70 in monthly credits - **Copilot Max** ($100/month): $200 in monthly credits Model choice now directly affects cost. Prices range from $0.25/million tokens (GPT-5 mini) to $5.00/million tokens (GPT-5.5 or Claude Opus 4.8). Code completions and Next Edit suggestions are not metered — only chat, agent mode, multi-step sessions, inline assists, commit-message generation, tool calls, and subagent work. For business and enterprise tiers, credits pool at the org level rather than per-user buckets. There's a layered budget control system: user-level budgets, cost center budgets, and enterprise-wide caps. Worth noting: the enterprise cap only limits metered overage charges after the shared pool is exhausted, not total spend. ## Why the old model was unsustainable The previous structure was, in retrospect, absurd. As Theo put it: you could do $40,000+ of inference for $40. The new system requires at least $60 of inference to hit the $40 plan limit — which is actually a more defensible ratio, even if it stings for heavy users. This isn't unique to GitHub. AI products across the board have been heavily subsidized by investor money. Uber reportedly burned through its entire 2026 AI coding budget by April. The pattern is consistent: developers using company-paid accounts don't feel the cost, so they don't moderate usage. One developer spent $1,000 in a single day via API and described it as "practically free" because their employer was paying. Autonomous agents make this worse. Agentic loops use roughly 30x more tokens than regular chat to complete the same work. When you're paying per token, that math gets uncomfortable fast. ## Real-world cost examples The numbers aren't theoretical. One developer reported that a single agentic prompt using Claude Sonnet 5 on max effort consumed 9% of their monthly Pro+ credit allowance in one session. Other reports show costs jumping from $29 to $750/month, or $50 to $3,000/month for heavy agentic use. For context: antirez (Redis creator) ships meaningful work within a $200/month AI budget across OpenAI and Anthropic combined. Heavy spending isn't a prerequisite for productivity — but uncapped agentic workflows can get there fast. ## The governance gap Copilot's model controls currently operate at the organization level, which is too coarse now that model selection is a financial decision. A developer choosing Claude Opus over GPT-5 mini isn't just picking a UX preference — they're making a spending decision that affects the whole org's credit pool. What's missing: granular model permissions by team, role, or user, similar to how cloud platforms manage resource access. Ideally, a policy-aware model router would automatically direct tasks to appropriate models based on organizational rules, rather than leaving every developer to choose manually. Microsoft's Foundry model router does something like this; Copilot doesn't yet. In the meantime, practical steps for enterprise admins: - Set hard spending caps at the user and cost-center level before the shared pool runs out - Restrict access to the most expensive models for general use cases - Use the new `ai_credits_used` field in the Copilot usage metrics API to track consumption per user per day - Audit which workflows are actually driving token spend — agentic sessions are almost certainly the culprit ## Alternatives worth considering Some users are switching. Claude Code (Anthropic's own CLI tool) offers comparable output quality at $20/month baseline, with a $200/month tier for heavy use. One developer ran both tools on the same Dapr workflow task and found similar output quality, with Claude Code noticeably faster. Zed users on Copilot Chat can switch to alternative providers: ChatGPT, OpenCode, custom API keys, or local models — none of which are affected by the credit changes. For teams that want to avoid single-vendor lock-in entirely, BYOK (bring your own key) setups with multi-model routing give more granular cost control than any managed plan currently offers. ## The bigger picture This is probably the right direction, even if the transition is painful. Subsidized AI usage was never going to last — as AI companies approach IPOs and investor patience thins, costs have to land somewhere. Token-based billing at least makes the economics visible. The uncomfortable truth is that most organizations don't actually know what their AI usage costs at real rates. Now they're finding out. ## Community discussion Top comments from developers on daily.dev. **@bennamanemourad** · 0 upvotes > Interessant ! ## Similar posts on daily.dev - [GitHub moves Copilot to usage-based billing as AI coding costs climb](https://daily.dev/posts/github-moves-copilot-to-usage-based-billing-as-ai-coding-costs-climb-angxve7kl) · The New Stack · 1 upvotes · 0 comments --- Tags: [#github](https://daily.dev/tags/github), [#ai-coding](https://daily.dev/tags/ai-coding), [#zed](https://daily.dev/tags/zed) [View this post on daily.dev](https://daily.dev/posts/github-copilot-switches-to-token-based-billing-what-changes-and-what-doesn-t-dqgznlmcl) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"GitHub Copilot switches to token-based billing: what changes and what doesn't","url":"https://daily.dev/posts/github-copilot-switches-to-token-based-billing-what-changes-and-what-doesn-t-dqgznlmcl","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/github-copilot-switches-to-token-based-billing-what-changes-and-what-doesn-t-dqgznlmcl"},"datePublished":"2026-06-02T15:11:30.511Z","dateModified":"2026-06-19T16:32:03.260Z","description":"GitHub retired its premium request model on June 1, replacing it with token-based 'AI Credits' billing. Plan prices remain unchanged, but usage is now metered...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/6329afbee65c5655d1093dd5d77db946?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/6329afbee65c5655d1093dd5d77db946?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"Collections","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"Collections","logo":"https://media.daily.dev/image/upload/s--fk_6ycEi--/f_auto,q_auto/v1780996001/logos/collections?_a=BAMAMiWQ0","url":"https://daily.dev/sources/collections"},"commentCount":2,"discussionUrl":"https://daily.dev/posts/github-copilot-switches-to-token-based-billing-what-changes-and-what-doesn-t-dqgznlmcl","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":3},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":2}],"keywords":"github,ai-coding,zed","timeRequired":"PT4M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"Collections","item":"https://daily.dev/sources/collections"},{"@type":"ListItem","position":3,"name":"GitHub Copilot switches to token-based billing: what changes and what doesn't"}]} {"@context":"https://schema.org","@type":"WebPage","@id":"https://daily.dev/posts/github-copilot-switches-to-token-based-billing-what-changes-and-what-doesn-t-dqgznlmcl","comment":[{"@type":"Comment","text":"Interessant !","datePublished":"2026-06-03T07:33:31.830Z","url":"https://daily.dev/posts/DqgZnLMCL#c-mAq8XO503","author":{"@type":"Person","name":"Bennamane Mourad","url":"https://daily.dev/bennamanemourad","image":"https://avatars.githubusercontent.com/u/90926833?v=4"}}]} ```