AI Pricing Crisis: GitHub Copilot Ends Flat-Rate, Industry Faces Cost Scramble
Key Questions
Why did GitHub Copilot switch from flat-rate pricing?
GitHub Copilot moved to usage-based credits starting June 1, leading to reported cost increases of up to 25x for some users. This change has triggered industry-wide backlash and prompted companies like Uber and Priceline to face significant budget overruns.
What is the Tokenomics Foundation and why was it formed?
The Tokenomics Foundation was established in response to the AI pricing crisis affecting multiple organizations. It aims to address the challenges of rising AI API costs across the industry.
How does GLM-5.2 compare to other models like GPT-5.5?
GLM-5.2 offers open weights, a 1M context window, and costs 6x less than GPT-5.5 while delivering 2x faster throughput on Vercel. It has been noted as potentially leading in frontend coding tasks following certain model removals.
What does the Gartner report predict about AI coding costs?
The Gartner report warns that AI coding tool expenses could exceed average developer salaries by 2028. This reflects broader concerns about total AI spend rising despite falling per-unit prices due to factors like communication overhead in agents.
Which companies have shifted to usage-based AI pricing?
Both GitHub Copilot and Microsoft Copilot have transitioned to usage-based models. A separate industry report indicates that 87% of AI sellers plan similar pricing adjustments within the next 12-18 months.
GitHub Copilot's switch to usage-based credits (June 1) has caused backlash with 25x cost increases. Industry-wide analysis reveals Uber, Priceline facing budget blowouts, and the formation of the Tokenomics Foundation. DeepSeek V4 Pro beats GPT-5.5 Pro in precision and costs 5% of Claude. DeepSeek now adds vision capabilities. GLM-5.2 released with open weights, 1M context, and 6x lower cost than GPT-5.5, now available on Vercel with 2x faster throughput. A tweet claims GLM-5.2 (Max) is now #1 for frontend coding after Fable's removal. GLM-5.2's official IDE ZCode launched with 1.5x usage quota for subscribers and BYOK support. A broad AI pricing report indicates 87% of sellers plan pricing changes in 12-18 months. Microsoft Copilot has also moved to usage-based pricing. A Gartner report warns AI coding costs could exceed developer salaries by 2028. A new analysis reveals that falling unit prices are masking rising total bills due to 'communication tax' in agents. OpenAI has started placing ads on paid plans. Claude has 13% paying users vs ChatGPT's 8% and 1.5x revenue per user. Software developer rates have risen 34-44% since AI went mainstream. New major model launches: GPT-5.6 Sol (record agentic benchmarks) and Sonnet 5 (adaptive thinking, tokenizer change as cost trap) were released in one day, intensifying the agentic race and adding pressure on pricing models.