AI costs how much? GitHub Copilot users react to new usage-based pricing system.

2026-06-01 · Source: AI - Ars Technica · Field: Technology & Digital — Software Development & Engineering, Artificial Intelligence & Machine Learning · Depth: Intermediate, short

Summary

GitHub Copilot transitioned from request-based to usage-based billing in April, with the new system taking effect today. This change has caused significant user sticker shock, as many report quickly exhausting their monthly AI credits. Previously, GitHub subsidized power users, with some estimating their old usage would now cost thousands of dollars. The new model grants monthly AI credits, where one credit equals \$0.01 of usage. For instance, the \$10/month Pro plan includes 1,500 credits (\$15 worth). Credit consumption depends on input/output tokens and the underlying large language model, with GPT-5.4 nano costing \$1.25 per million output tokens versus \$30 for GPT-5.5. Users relying on "Auto" mode risk incurring high costs, as it can select expensive models for simple queries. Some users have consumed thousands of credits in hours, while others are adapting by using AI more deliberately.

Key takeaway

For AI Engineers or Software Developers relying on AI coding assistants, your usage patterns must adapt to new usage-based billing models. If you continue with previous "power user" habits, expect significantly higher costs, potentially thousands of dollars monthly. You should actively monitor your token consumption, manually select cost-effective LLMs, and be deliberate in your AI interactions, especially by limiting chat context to avoid unnecessary input token charges. This shift necessitates a more strategic approach to AI tool integration.

Key insights

GitHub Copilot's shift to usage-based billing reveals significant AI inference costs previously absorbed by providers.

Principles

• AI inference costs scale with token usage and model choice.
• "Auto" model selection can lead to unexpected high expenses.
• Deliberate AI interaction reduces credit consumption.

In practice

• Monitor token usage for AI coding assistants.
• Manually select cost-efficient LLMs for tasks.
• Limit chat context to reduce input token costs.

Topics

• GitHub Copilot
• Usage-Based Billing
• AI Inference Costs
• Large Language Models
• Code Generation
• Software Development

Code references

• orgs/community

Best for: CTO, VP of Engineering/Data, Machine Learning Engineer, Software Engineer, AI Engineer, Director of AI/ML

Related on AIssential

• GitHub will start charging Copilot users based on their actual AI usage — AI service providers are shifting to usage-based billing to manage escalating inference costs and ensure sustainability.
• Microsoft's GitHub shifts to metered AI billing amid cost crisis — GitHub Copilot is shifting to usage-based billing to manage escalating AI inference costs.
• ‘What a joke’: Github Copilot’s new token-based billing spurs consternation among devs — GitHub Copilot's new token-based billing model, effective June 1, significantly raises costs for many users, prompting debate over usage efficiency and Microsoft's role.
• Inside the Forward Pass: Pre-Fill, Decode, and the GPU Economics of Serving Large Language Models — LLM economics are shifting from one-time training costs to perpetual inference expenses due to massive token consumption.
• Your AI Pricing Problems Are Actually Business Model Problems — AI's usage-based costs fundamentally break traditional SaaS unit economics, demanding a business model transformation.
• Changes to GitHub Copilot Individual plans — Agentic AI workflows significantly increase compute demands, necessitating pricing model adjustments.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by AI - Ars Technica.