Recently, everyone seems to feel that Tokens are becoming increasingly expensive, and the limits provided by various companies are becoming less sufficient.

Thus, how to save money and use more Tokens has become a frequently discussed topic among everyone.

I often see many friends in various groups discussing where to get free Tokens, for example, WorkBuddy provides 100 points daily, and DeepSeek V4.1 Flash can be used for free for a long time. Everyone frequently reminds each other to collect points today and often shares interesting benefits.

This indeed seems to be a necessity for everyone.

I spent a few days creating a little feature that I think some friends might like, which is:

Token Benefits.

In the past, when everyone wanted to know about some Token benefits, it was quite troublesome because the information was too scattered.

Some was on official websites, some on public account messages, and some were posted on X, etc.

Moreover, these Token benefits have a time limit.

For example, if you used the newly launched search API from DeepSeek in some Agents, it was actually free, but free until November 1, so it ended a few days ago.

So finding Token benefits is essentially not a technical difficulty; it is basically a matter of collecting information + processing and analyzing information.

And this is exactly the task that AIHOT does every day.

Some friends might not be very familiar with AIHOT, so I’ll introduce it briefly.

AIHOT is an AI news site we built ourselves, aihot.news, which now has over 1 million monthly active users and has open-sourced its underlying engine and architecture, currently garnering nearly 10,000 Stars.

It now monitors nearly thousands of information sources, with AI automatically analyzing and picking out truly worthwhile AI updates from the entire AI world every day, creating curated recommendations and daily/weekly updates based on a 5-minute monitoring dimension. If you want AI news, AIHOT is sufficient.

Additionally, we have also opened free and anonymous Skill, MCP, and API, allowing everyone to freely connect with Agents and for secondary creation use.

Since there is an overwhelming amount of information every day, there are certainly countless duplicates. If everything is pushed out, it will flood the screen, leading to a poor user experience. Moreover, the number of sources that report an event is itself a reflection of its popularity. Therefore, I developed a complete set of clustering algorithms to categorize events, pushing the most representative and core news while other news turns into related messages. I also wrote a heat algorithm, thus creating a hot list.

However, with so many sources and pieces of information being monitored and stored in my database daily, the potential is far beyond just reading news.

Long ago, many people were actively exchanging news about Tibo releasing the Codex reset, wanting to know when Codex would reset. Because AIHOT has been monitoring Tibo’s X, we have been monitoring and storing every statement from Tibo in our database in real time, transforming the originally monitored information into a Codex reset monitoring feature was quite simple for us.

Therefore, as soon as Tibo posts, we can immediately inform you at what time it takes effect Beijing time, who can use it, and what his exact words were, while also creating a calendar to mark every reset.

Recently, everyone has mentioned that a guy named Andrew will be taking over from Tibo. We have been monitoring this person as well and additionally monitored him while also extending the monitoring to Google’s Antigravity and Claude reset.

This, in fact, is an auxiliary product created after processing the many sources and structured information that AIHOT monitored in the past, turning it into something beneficial for everyone.

Then a while ago, people started discussing how Tokens were becoming more expensive, and I suddenly realized that these benefits are, in essence, information.

I often see many vendors announcing such benefits, new users getting a certain amount, daily sign-ins granting points, and which models are free for a limited time.

However, in the previous AIHOT logic, all of these things had to be filtered out.

Because from a news perspective, they are just advertisements, just noise, and should absolutely not appear in our curated list.

I still remember the effort I put into creating filtering rules to eliminate these ads, but looking back now, isn’t this exactly what many people are looking for in groups every day?

To news, it is noise.

But from the demand side, it is tangible benefits.

Since this is the case, let’s do the opposite.

By reversing those filtering rules, isn’t it looking for “what kind of information is a benefit announcement”?

Thus, this idea emerged, and hence we have today’s Token Benefits.

Here is the link: aihot.news/free-tokens

When you open it, the largest number in the upper left corner indicates how much you can still receive today. When I wrote this article, it showed 201 dollars.

The two smaller numbers next to it indicate how much you have collected this month and how much time is left until the next refresh.

Below, all benefits are divided into several groups based on the collection rhythm: daily, weekly, monthly, and other benefits, such as new user packages, limited-time free offerings, and free models under certain conditions.

You can directly click on the right to collect. Some benefits need to be collected within their clients. For example, for WorkBuddy, the daily points need to be collected by signing in to their client.

So, to enhance everyone's user experience on computers, I developed a shortcut to easily launch the client. After clicking this button, if you have the client installed, the browser will directly ask you whether to open it, and you can simply allow it, and the client will open, and you can collect right away. If you haven’t installed it, we will directly help you open its official download page.

With this step completed, a usable Token benefits collection has been established.

However, I still feel that something is not particularly engaging.

Because collecting benefits can indeed be a bit monotonous; opening a few clients daily, clicking to sign in, collecting 100 points, and then nothing follows, and I cannot develop a direct Agent to assist everyone because the compatibility and differences among various systems and agents are too great.

So how can we make this process feel enjoyable? I want everyone to feel a sense of achievement while collecting Token benefits, to intuitively see how much money they are saving and how much they have accumulated.

It is essentially providing everyone with a bit of emotional value.

Then, I remembered the card-based mobile games I enjoy, leading to the fun idea of making it into a card collection album...

Now, each benefit on the page is represented as a card. The ones not collected are ordinary faint cards, and once collected, they will be illuminated.

Then, the cards will become brighter the more you collect.

For those collectible daily, if you accumulate for 7 days, it becomes a blue card, 15 days a gold card, and 30 days a flash card. For weekly collections, it’s 2 times, 4 times, and 8 times; for monthly collections, it’s once, twice, and thrice. The once collectible ones depend on how valuable your collection is: starting at 10 dollars is a blue card, 50 dollars a gold card, and 100 dollars directly a flash card.

I personally really like this design = =

The flash cards have a dazzling design, with a rotating rainbow border, shimmering lights on the card face; when you hover the mouse over it, it enlarges and tilts with your mouse movement, providing ample emotional value...

In this entire card album, the brighter cards are placed at the front; flash cards are at the very front, followed by gold cards, blue cards, and uncollected ones at the back. The collectable once cards are valued based on their worth at the moment you collected them; even if the vendor changes the price or conversion rate later, the cards you have will neither increase nor decrease in value.

If certain Token activities have ended, the card you collected will not disappear.

It will be stamped with a red "Limited Edition" stamp, indicating its last day, while retaining the level you accumulated, placed at the end of the card album; uncollected ones will not be kept.

The activity may be gone, but the cards will remain, serving as a keepsake for everyone.

Similarly, in the list, once collected, that row will be stamped with a mark saying "Collected today" along with the date. If you clicked incorrectly, you can hover over the stamp and undo it.

Honestly, these features are not significant functions.

But I believe that if collecting Token benefits can make everyone a little happier each day, while also providing some emotional value when receiving benefits, then that would be wonderful.

Finally, I’ll briefly talk about the technology used for Token benefits.

The first layer is the radar.

In addition to the original news information sources of AIHOT, we specifically added a batch of sources for Token benefits, including vendors' accounts and pages, major players specifically digging for benefits, and various collections from the community.

The radar reads the new content from these sources every hour, first passing it through a deliberately wide keyword net, covering terms like free, grab, hedge, benefits, limits, bonuses, sign-in, and limited free, including the English terms free and credits; this layer is sufficiently broad, preferring to include more.

After all, adding one more entry only means spending a very small and short model call, but missing one will never be seen again.

After this layer of preliminary screening, the content is passed to a larger model for an additional layer of certainty judgment, listing every activity mentioned, judging whether it qualifies as a benefit and explaining why it does not qualify, whether it’s a transit station, a rebate, or just for paying users.

The second layer is verification.

What the radar finds is merely a clue. It will only summon an agent for verification if the vendor's own account confirms it, if two different sources are saying it, or if the post directly links to the vendor’s own page.

If it’s just a solitary community post, it will be left pending until the second source appears.

When the agent verifies, it only recognizes the vendor's own page and must confirm that this page truly belongs to the vendor since there are indeed many fake promotional pages online, so this verification agent must navigate from the vendor’s homepage or its official account to this activity page to confirm its authenticity.

After verification, it must pass an additional hurdle called "worth collecting." It can only use very old models like GLM-4.7 or DeepSeek v3; otherwise, it’s certainly not accepted because it lacks meaningfulness.

Many marketing articles like to aggregate "each model giving away 1 million Tokens" into "giving away 70 million Tokens"; this type will be directly rejected by the verification agent. Various rebate links, API transit stations, secondary points, shared accounts, and receiving codes for multiple logins will also be rejected, as will those requiring payment before receiving, since people cannot claim them.

Honestly, these are all hard-earned lessons from previous experiences...

The third layer is inspection.

Once benefits are launched, the vendor can change the rules at any time, stop early, or the limits may have already been fully claimed. Therefore, when each item is launched, we copy one to three sentences from the vendor's page, detailing how much and how to claim it, and until what day; a script is specifically run daily to verify this twice.

If the original wording cannot be found for two consecutive times, or if new words like "paused" or "ended" appear on the page, an agent is immediately dispatched for re-inspection.

Then, each night, an agent follows a checklist to review every Token benefit listed on the page, and the next morning, newly launched or changed links will be compiled into a list for me and another agent to confirm.

This entire set is mostly automated by the agents; each morning, I just need to scan through that list like reviewing documents.

Overall, that’s how it works.

Finally, I hope this little feature can be of some use to everyone.

Honestly, I have always felt that the world of AI is unequal.

Everyone knows that if you really want to use a better model for work, run Agents, and call APIs, it almost always requires money.

But I have also seen many people wanting to keep up with this era, wanting to try out what these large models can do, wanting to use Agents for interesting things, but in the end, they are blocked by limits and costs.

I also understand how hard the first step is because that’s real money.

Thus, I created this little feature.

It does not touch any gray areas, using only publicly available operating activities from various vendors. Vendors sending out benefits are naturally hoping that more people will use and try their services. What we are doing is aggregating these scattered activities into one page so that more people in need can see them.

Of course, to be frank, the behavior of collecting Token benefits is certainly not a long-term solution; activities will always end, and free limits will eventually run out.

But I hope it allows you to open the door without spending a dime and start using it.

Once you are comfortable using it and feel it’s valuable, then you will naturally weigh the pros and cons when you spend money.

In conclusion, today’s Token Benefits have just gone live, and there are certainly many areas that are still imperfect.

If you see benefits that we have not included, you can click on “Report a benefit” at the top of the page to inform us about it, or any errors as well.

Each limit shown on this page might not be as abundant as stated.

However, I hope that for many people, their first real experience using AI starts from here.

I also hope to do my best to help everyone smooth out some of those information gaps.

Allowing everyone to feel the charm of AI.

I also hope to spark curiosity about AI in everyone.

That said, since you've read this far, if you think it’s good, feel free to like, follow, and share. If you want to receive notifications first, you can give me a star mark⭐~ Thank you for reading my article, and I look forward to seeing you next time.

>/ Author: Kazk

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到[email protected]，本平台相关工作人员将会进行核查。