OpenCode's 70-Yuan Subscription Delivers 3.6 Billion Tokens, Slashing AI Costs Tenfold

Hello everyone, I'm a rural programmer (returned to the countryside in 2020, haven't left the village since), an independent developer (committed to developing 1,000 small products), and the front-end tiger Chen Suiyi (self-appointed). My official site: https://yicode.tech.

Today we're not talking about making money; we're talking about saving money.

Look at the data directly: in the last 10 days, I used 3.6 billion Tokens, and my 70-yuan subscription plan hasn't even run out.

Comparing with Deepseek official, 160 million Tokens costs 10 yuan, so 3.6 billion Tokens would cost over 200 yuan.

Yesterday I went hard and consumed nearly 1 billion Tokens.

Note that my 70-yuan subscription plan is only 80% used, with 20% left.

Roughly estimating, a full 70-yuan plan can consume around 5 billion Tokens, which is equivalent to about 300 yuan at Deepseek's official pricing.

That means, with the same Deepseek v4.1 flash model, a 70-yuan subscription can unlock 300 yuan worth of usage.

Before this, I spent close to a thousand yuan per month on AI; that hurt, a lot.

With my current usage, buying 3 subscription plans for 200 yuan is more than enough for smooth use, which is unbelievable.

This platform is called opencode, official site: https://opencode.ai.

As shown above, DeepSeek V4.1 Flash is the lowest-consuming tier, and the basic plan allows 34,000+ requests, insane!

But I'm puzzled: why is the Plus plan 4 times the price of the Base plan, yet only 2 times the usage?

Wouldn't buying 2 Base plans be equivalent to one Plus plan? And it's 140 yuan cheaper. This pricing design is beyond my understanding.

But as long as I can use it smoothly, I don't care about these details. The Deepseek v4.1 flash model's capability is top-tier, and with this budget plan I'm sharing, even relay stations can't compete.

Especially convenient, opencode supports WeChat and Alipay payments, so those without credit cards don't need to worry.

If you find this article useful, please give it a one-click triple like. Thanks, bros.