Why Has Credit Consumption Increased 4x Compared to Three Months Ago?

[HeyItsDizzy]

🏷️ Discussion Type

Bug

💬 Feature/Topic Area

Copilot in GitHub

Body

I've noticed a significant increase in credit consumption over the past few months, and I believe there is now a serious mismatch between the complexity of tasks being requested and the credits being charged.

I appreciate the improvements in reasoning, planning, and overall AI capability. However, the model selection and token usage seem increasingly disconnected from the actual task being performed.

For example, changing the background colour of a text input should not consume 50–100 credits, especially when using "Low" reasoning. I understand that AI convenience has a cost, and I am not arguing that it should be free. What I am saying is that the result needs to be proportional to the credits charged.

A recurring issue appears to be Auto mode. From my experience, Auto behaves as though it is frequently selecting High or Max reasoning models, even for simple UI changes. The result is that minor modifications can cost hundreds of credits.

Some recent examples:

Example 1

• Claude Sonnet 4.6
• 578.0 credits
• Approximately 100 lines changed

While this is not the worst example, I would have expected a task of this size to cost somewhere in the range of 150–200 credits based on historical pricing and usage patterns from only a few months ago.

Example 2

• Claude Sonnet 4.6
• 81.7 credits
• 13 lines changed (+5 / -8)

Followed immediately by:

• Claude Sonnet 4.6
• 89.9 credits
• 1 line correction

The second prompt required almost no reasoning. It was simply correcting a previous response. Yet it cost more than the original 13-line modification.

This is difficult to understand from a user perspective.

To put this into context, three months ago a 300–400 credit task would often involve generating an entire product catalogue section, linking categories, creating data structures, and performing meaningful architectural work. That felt fair.

Today, the same credit consumption is occurring on relatively minor UI tweaks and small corrections.

I currently have a 7,000-credit monthly allocation plus an additional $50 budget. I've consumed almost all of that within 10 days, despite my coding intensity remaining relatively unchanged.

I'm not coding full-time. This is a hobby project and quality-of-life application that I work on in my spare time. My workload and prompt complexity have not increased enough to explain a fourfold increase in costs.

My request is not for lower prices across the board. My request is for model selection and credit consumption to better reflect the complexity of the task being asked.

If I ask for:
"Change the background colour of this input from white to blue"

I should not be paying anywhere near the same amount as:
"Design and implement an entire product catalogue architecture."

The current pricing behaviour makes it difficult to predict costs and is causing me to question whether Auto model selection is functioning appropriately for simple development tasks.

I would appreciate some transparency around:

• Why costs have increased so dramatically.
• How Auto selects reasoning levels.
• Why simple correction prompts can cost more than larger implementation prompts.
• Whether there are plans to better align credit consumption with task complexity.

At the moment, the value received no longer feels proportional to the credits being consumed.

[github-actions[bot]]

💬 Your Product Feedback Has Been Submitted 🎉

Thank you for taking the time to share your insights with us! Your feedback is invaluable as we build a better GitHub experience for all our users.

Here's what you can expect moving forward ⏩

• Your input will be carefully reviewed and cataloged by members of our product teams. 
  • Due to the high volume of submissions, we may not always be able to provide individual responses.
  • Rest assured, your feedback will help chart our course for product improvements.
• Other users may engage with your post, sharing their own perspectives or experiences. 
• GitHub staff may reach out for further clarification or insight. 
  • We may 'Answer' your discussion if there is a current solution, workaround, or roadmap/changelog post related to the feedback.

Where to look to see what's shipping 👀

• Read the Changelog for real-time updates on the latest GitHub features, enhancements, and calls for feedback.
• Explore our Product Roadmap, which details upcoming major releases and initiatives.

What you can do in the meantime 💻

• Upvote and comment on other user feedback Discussions that resonate with you.
• Add more information at any point! Useful details include: use cases, relevant labels, desired outcomes, and any accompanying screenshots.

As a member of the GitHub community, your participation is essential. While we can't promise that every suggestion will be implemented, we want to emphasize that your feedback is instrumental in guiding our decisions and priorities.

Thank you once again for your contribution to making GitHub even better! We're grateful for your ongoing support and collaboration in shaping the future of our platform. ⭐

[Zero-day-Exploit-np]

Description

I've noticed a significant increase in credit consumption over the past few months, and it now feels like there is a serious mismatch between the complexity of tasks and the credits being charged.

I want to be clear: I appreciate the improvements in reasoning, planning, and overall AI capability. I understand that convenience comes with a cost, and I am not asking for free usage. However, the pricing should feel proportional to the task being performed.

Right now, that balance seems off.

---

Observed Issue

The main concern appears to be Auto mode, which often behaves as if it's selecting High or Max reasoning models, even for very simple tasks.

For example:

Changing the background color of a text input should not cost 50–100 credits, especially when using "Low" reasoning.

Yet, this is happening frequently.

---

Examples

Example 1

• Model: Claude Sonnet 4.6
• Cost: 578.0 credits
• Changes: ~100 lines

While not extreme, I would have expected this to fall in the 150–200 credit range, based on usage patterns from a few months ago.

---

Example 2

• Model: Claude Sonnet 4.6
• Cost: 81.7 credits
• Changes: 13 lines (+5 / -8)

Followed immediately by:

• Model: Claude Sonnet 4.6
• Cost: 89.9 credits
• Changes: 1 line correction

The second prompt required almost no reasoning, yet it cost more than the original change, which is difficult to understand from a user perspective.

---

Comparison with Past Usage

Around 3 months ago:

• 300–400 credits would typically involve:
  • Generating full features
  • Designing structures
  • Implementing meaningful logic

That felt fair.

Today:

• Similar credit consumption is occurring on minor UI tweaks and small fixes

---

Personal Impact

• Monthly allocation: 7,000 credits + $50 budget
• Consumed almost entirely within 10 days
• Coding intensity: unchanged
• Usage type: hobby project (not full-time development)

The increase in cost does not match the increase in workload or prompt complexity.

---

Core Concern

This is not a request for lower pricing overall.

The core issue is:

Credit usage should better reflect task complexity.

Simple requests like:

"Change the background color of this input from white to blue"

Should not cost anywhere near:

"Design and implement a full product catalogue system."

---

Questions / Request for Transparency

I would appreciate clarification on the following:

• Why have costs increased so significantly?
• How does Auto mode determine reasoning level and model selection?
• Why do simple correction prompts sometimes cost more than larger implementation tasks?
• Are there plans to improve alignment between:
  • Task complexity
  • Model selection
  • Credit consumption?

---

Summary

At the moment, the value received no longer feels proportional to the credits being consumed.

Improving transparency and better aligning cost with task complexity would significantly improve the developer experience.