Unfair Pricing Model: Claude Opus 4.5 Should Not Cost 3x More Than Other Premium AI Models

[hamidehgb69-afk]

Select Topic Area

Question

Copilot Feature Area

Copilot in GitHub

Body

Dear GitHub Copilot Team,

I am writing to express my strong dissatisfaction with the current pricing structure for AI models in GitHub Copilot, specifically regarding Claude Opus 4.5 consuming 3x credits compared to other premium models that only consume 1x credits.
The Problem: Discriminatory Pricing

As a paying Copilot Pro subscriber, I find it fundamentally unfair that:

Claude Opus 4.5: 3x credits per request
GPT-5, GPT-5. 2, Gemini 3 Flash, Claude Sonnet 4.5: 1x credits per request

This means that if I use Claude Opus 4.5—which is often the most suitable model for complex tasks—my Pro subscription runs out 3 times faster than if I use other premium models. This is not a sustainable or equitable pricing model.
Why This Is Unacceptable

1. We Already Pay for Premium Access

Copilot Pro subscribers pay a monthly fee specifically to access advanced AI models. Charging 3x for one particular model effectively creates a "premium within premium" tier that punishes users for choosing the best tool for their work.
2. Other Platforms Don't Do This

Competitors like ChatGPT Plus and Claude Pro offer unlimited access to their top-tier models (GPT-4o, Claude Opus) for a flat monthly fee. GitHub Copilot's credit-based system with discriminatory multipliers is anti-consumer and anti-competitive.
3. It Defeats the Purpose of Multi-Model Selection

If one model costs 3x more, users are effectively discouraged from using it, which contradicts the entire point of offering multiple models. This is not "choice"—it's forced downgrading.
4. It Disproportionately Affects Heavy Users

Developers who rely on Copilot for professional work—especially those working on complex projects that genuinely benefit from Claude Opus—are unfairly penalized for doing their jobs effectively.
What Needs to Change: Our Demands

We, as paying customers, respectfully but firmly request the following:
✅ Option 1: Make Claude Opus 2x Instead of 3x (Minimum Acceptable Change)

Reduce the credit cost of Claude Opus 4.5 from 3x to 2x. This would:

Still reflect its premium nature
Make it accessible for regular use
Prevent subscriptions from depleting too quickly
Show that GitHub values its Pro subscribers

✅ Option 2: Equalize All Premium Models at 1x (Ideal Solution)

Treat all top-tier models (Claude Opus, GPT-5, Gemini 3 Flash) equally at 1x credits. If a model is offered in Copilot Pro, it should be fully accessible to Pro subscribers without arbitrary penalties.
✅ Option 3: Offer a "Pro+" Tier with Unlimited Claude Opus

If GitHub insists on treating Claude Opus differently, create a Pro+ tier (similar to competitors) that offers unlimited access to all models, including Opus, for a higher—but transparent—monthly fee.
The Bigger Picture: Trust and Fairness

This isn't just about pricing—it's about respecting your customers. We chose GitHub Copilot because we trust Microsoft and GitHub to provide fair, professional tools. But this 3x pricing model feels like:

❌ A bait-and-switch: "Here's a great model... but you can barely use it."
❌ Nickel-and-diming: Making us pay more for features already included in our subscription.
❌ Undermining Pro subscriptions: What's the point of paying for Pro if the best model is effectively locked behind a paywall-within-a-paywall?

Conclusion: Do the Right Thing

We are not asking for free services. We are asking for fair treatment as paying customers.

Reduce Claude Opus from 3x to 2x credits at minimum. This simple change would:

✅ Show you listen to user feedback
✅ Make Pro subscriptions sustainable for heavy users
✅ Restore trust in GitHub Copilot's value proposition
✅ Keep GitHub competitive with other AI platforms

We love GitHub Copilot. We want to continue using it. But this pricing model is pushing loyal customers away. Please reconsider this decision before it's too late.

Thank you for taking the time to read this, and I sincerely hope you will act on this feedback.

Respectfully but urgently,
A Concerned Copilot Pro Subscriber

[AlexanderJohnston]

Opus 4.5 and GPT 5.2 both have different token input/output costs. A glance at Opus shows that it is roughly 3x the cost.

GPT-5.2 Cost
$1.75 million_token/input
$14 million_token/output

Opus 4.5 Cost
$5 million_token/input
$25 million_token/output

GPT-5.2
250k input = $0.438
100k output = $1.4
= $1.839 request cost

Opus-4.5
250k input = $1.25
100k output = $2.5
= $3.75 request cost

Ratio
$3.75 / $1.839 = 2.04 <--- THIS exact ratio varies depending on the ratio of input and output tokens. It can go higher.

Since most requests will be between 2.0x and 3.0x of GPT-5.2's cost, Opus-4.5 should cost 3x base.

[gpswd]

And to add onto this, Anthropic is the only one that charges extra for input cache writes. The other providers do it for free and implicitly, Anthropic doesn't and has a 5 minute expiry.

Opus 4.5 cache hits are $0.50/mTok.
GPT 5.2 is $0.175 (2.8x less).
3.0 Pro is $0.20 (2.5x less).

And that's before factoring in the $1.75 extra on top of the $5 default rate Anthropic charges to cache in the first place.

So make sure you never go over 5 minutes between prompts if using Opus 4.5, you'll get a lot more usage out of it that way.

[augusto-dmh]

Soon we hope to not have to pay anything for cloud-based models and instead have optimized models in our local computer.

That will be the game changer.

[Thaval]

Yea, I was hoping for Mistral as it was planned to have open source and free but guess what... nobody in the AI business can't get enough.

[gpswd]

Anthropic has never offered an unlimited Opus usage plan.

[Thaval]

I'd also love to have lower costs for Opus 4.5 but that's not something Github can be blamed for, they just forward the costs.