Can I ask? What is the benefit for me to switch from Pro to Pro+ plan? Isn’t it that I will simply use up the $60 just as quickly as if I were using usage-based pricing? Maybe I don’t understand the new payment model yet, so thanks for the explanation.
1 Like
Tried both, and I feel that’s exactly it. The only difference I see is that with usage based pricing there is more transparency.
In addition to the $60 API that you are guaranteed, you also receive an API quota that is updated every 5-24 hours. The Pro+ quota is three times higher.
They removed this and there is no more recharge of schedules
What do you mean? It’s 3x higher because it’s 3x more expensive, right?
Do they keep changing the rules every week?! 0_o
It looks like this… for me as a developer who uses Cursor quite a lot every day it’s quite problematic 
If what @Artemonim wrote is removed now, there’s really no reason to subscribe to Cursor for those using Claude only vs subscribing directly to Claude Code where you do get more for less money.
I had GPT compare the pricing of claude code and cursor on claude-4-sonnet and it said it was the same. He wrote exactly: Cursor uses prices that match the new pricing from Anthropic.
In practice, you pay around $0.10–$0.30 per half a million tokens, which corresponds to a typical mix of input and output during normal usage.
Question is : why API is why does Api cost so much money? I mean the price for tokens.
Let’s do it wright with Gemini
1. Estimated API Cost Per Million Tokens
To estimate the costs, I used the usage data from your screenshot and referenced publicly available API pricing to establish a realistic ratio between input and output token costs. The calculation is based on the formula: Total Cost = (Input Tokens * Price_per_Input_Token) + (Output Tokens * Price_per_Output_Token).
The resulting estimates for cost per million tokens (CPM) are as follows:
| Model | Est. Input CPM ($/M tokens) | Est. Output CPM ($/M tokens) |
|---|---|---|
| Grok-4 | ~$3.11 | ~$15.57 |
| Gemini 2.5 Pro | ~$3.98 | ~$19.90 |
| Claude 4 (Sonnet) | ~$4.30 | ~$21.50 |
| Claude 4 Thinking | ~$7.49 | ~$37.45 |
| o3-pro | ~$28.85 | ~$86.54 |
| o4-mini | ~$0.17 | ~$0.68 |
| auto | ~$3.09 | ~$6.18 |
Note: o3 was omitted due to its minimal usage and cost, making estimation unreliable. The auto model is listed for reference as per the prompt; in practice, it is free for users. The costs for Claude 4 and Grok-4 are notably high in the provided data compared to their standard API rates of $3/$15 per million tokens, suggesting the specific workload involved heavy, complex requests.
2. Usage & Cost-Effectiveness Analysis
Analysis of Usage and Spending
Your concern about depleting your $60 package is valid. Having purchased it on July 7th, you have used it for 9 days as of July 16th. If the package runs out today, your average daily cost is approximately $6.67 per day.
- Projected Monthly Cost: At this rate, your projected 30-day spending would be over $200, far exceeding your $60 package.
- Cost Drivers: The provided data shows that high-end models like
grok-4-0709($50.10),gemini-2.5-pro-preview-05-06($34.16), andclaude-4-sonnet($23.77) account for the vast majority of expenses. Their powerful capabilities come at a significant cost.
Recommendations for Cost-Effectiveness
Once your package is depleted, you will move to a pay-per-use model, making cost management critical. To maximize your value:
- Prioritize Low-Cost Models: For standard or less complex tasks, shift your usage to significantly cheaper alternatives. The
o4-minimodel is the most economical choice according to the data, offering a very low cost per token. - Use Premium Models Strategically: Reserve expensive models like
grok-4,gemini-2.5-pro, andclaude-4-sonnetonly for tasks that absolutely require their advanced reasoning and generation capabilities. - Monitor Your Transition: Be prepared for the switch to pay-per-use. Continuing your current usage pattern will lead to high, direct costs. By adjusting your model selection, you can substantially reduce your spending.
3. Guide vs. Reality Assessment
The recommendations in the provided How To Optimize Your Usage: The Best AI Models to Use, version 2 are largely consistent with the cost analysis derived from the screenshot.
-
Agreement:
- The guide correctly identifies Auto as the free, default choice and o4-mini as a cheap and effective model for simpler tasks. The guide’s claim that
o4-miniis about “3x cheaper than Gemini 2.5 Pro or Claude 4” is a significant understatement; based on the data, it is dramatically cheaper—by a factor of more than 20x. - It positions Grok-4 as a premium, powerful, and expensive option, which the high API cost in the table ($50.10 for the workload) supports.
- The guide’s strategies (Budget, Balanced, Premium) correctly align with the cost hierarchy of the models.
- The guide correctly identifies Auto as the free, default choice and o4-mini as a cheap and effective model for simpler tasks. The guide’s claim that
-
Discrepancies:
- The guide considers Gemini 2.5 Pro the “former king” and now second to
Grok-4. Cost-wise, the analysis shows them in a similar premium bracket, making the choice between them more dependent on specific task performance than on price alone. - The guide’s dismissal of Claude 4 seems based on personal preference (“I’m not a huge fan of how proactive Claude can be”). From a cost perspective, it is positioned as a mid-to-high-tier option, making it a reasonable choice for the “Balanced Strategy” as the guide itself suggests.
- The warning about o3-Pro’s minimum cost is validated by the extremely high estimated CPM. The data strongly supports the guide’s caution to use it only when absolutely necessary.
- The guide considers Gemini 2.5 Pro the “former king” and now second to
no more quotas man, just “you have reached you limit”
Recalculated API Cost Per Million Tokens
The new estimates for cost per million tokens (CPM), factoring in the cost of Cache Read, are as follows:
| Model | Est. Input CPM ($/M tokens) | Est. Output CPM ($/M tokens) |
|---|---|---|
| Grok-4 | ~$3.21 | ~$16.05 |
| Gemini 2.5 Pro | ~$2.26 | ~$6.78 |
| Claude 4 (Sonnet) | ~$4.89 | ~$24.45 |
| Claude 4 Thinking | ~$3.92 | ~$19.60 |
| o3-pro | ~$25.35 | ~$76.05 |
| o4-mini | ~$1.49 | ~$5.96 |
| auto | ~$2.54 | ~$5.08 |
- Note on Calculation: These estimates are derived by treating
Cache Readas a billable event costing 10% of the standard input price. A common industry price ratio between output and input tokens (e.g., 5:1 for Claude, 3:1 for Gemini) was assumed to solve for the individual rates. The models with a high volume ofCache Readtokens (like Gemini and Claude) see a significant shift in their effective cost structure compared to the previous calculation.
No worries, you’ve just been charged $30 for nothing, and your “unlimited” plan will last only 10 days of regular use.) With such pricing policies, don’t even think about buying an annual subscription. Otherwise, it’ll run out in a month, and they’ll force you to keep buying more and more over and over again.)))