🎬 Pollo 2.5 is here! ⚡ What changed from 1.6 to 2.0 to 2.5? Testing the latest Pollo model under the same conditions! [PR]

Hello, this is Haru📘

Once again, this is brought to you as a joint writing project (PR article) with Pollo AI🎬

On Friday, December 26, 2025, Pollo AI officially released "Pollo 2.5"!

Pollo 2.5 is the latest model, equipped with two modes: Basic (high-speed generation) and Pro (cinema quality), which has significantly improved quality and efficiency compared to previous models.

This time, we will verify what has changed compared to the previously used Pollo 1.6 and Pollo 2.0, while comparing them under the same conditions!

🎥 First of all, what is Pollo AI?

Pollo AI is an
all-in-one generation tool that allows you to use cutting-edge AI models from around the world in one place.

You can try all of these on a single screen.

🔗 Pollo AI is explained in detail in this article👇

🔎 So, what kind of models were Pollo 1.6 and Pollo 2.0?

Now, I will explain the various Pollo models that have evolved so far!

⚡ Pollo 1.6 "First, get it running fast, the foundation for mass production"

Pollo 1.6 was a model with strengths in creating videos cheaply and quickly. It is convenient when you want to quickly run prototypes, and
being able to easily keep credit consumption low is also a charm.

On the other hand, the generated video is video only, and does not include sound (BGM or sound effects).

🎬 Pollo 2.0: "The visuals have improved, sound has been added, and the range of expression has expanded"

Pollo 2.0 retains the speed and cost-effectiveness of Pollo 1.6, while adding the ability to generate video with sound, bringing the final output much closer to a professional level.

Furthermore, a feature to naturally connect and extend videos has been added, making it easier to assemble even short clips into a single piece with a narrative.

In addition, the generation speed feels significantly faster, evolving into a practical model that allows you to quickly cycle through idea -> prototype -> fine-tuning.

Pollo 2.0 is also explained in this article here👇

🚀 So, what has changed with Pollo 2.5?

To put it simply, the changes in Pollo 2.5 are straightforward.

Compared to previous models, both quality and efficiency have increased significantly...!!

To verify that evolution...

🎰 I will compare the 3 models by creating a commercial video using the same prompt!

The theme for this verification is this👇

It is a
desk-resident idea gadget that gives you one topic for the note you should write today with just a single turn🎰

Let's turn this gadget into a video in the order of Pollo 1.6 ⇒ 2.0 ⇒ 2.5!👍

First, I used nanobanana to generate a still image of
me, Haru, trying to turn the "Idea Gacha Machine" while writing a note.

Here is the generated image👇

Click here for the prompt to generate the video👇
This time, to compare the differences in generation results, I will create them using
the same prompt for all 3 models.

<Video Prompt>

参照画像を最初のフレームとして使用。3Dカートゥーン調、暖かい室内照明、柔らかい陰影、滑らかな質感を維持。キャラクター、机、ライト、窓、背景、ガチャ本体の形と配置は一切変えない。
【最重要固定】
ガチャの操作部分は「参照画像と同じ金属の丸いノブ（つまみ）」を完全に維持する。
取っ手付きハンドル／クランク／レバー／棒状の持ち手を絶対に追加しない。ノブの形状変更・部品追加を禁止。【動き（連続）】キャラクターの右手が金属ノブに指先でしっかり触れ、ノブを時計回りにちょうど1回転（360度）ゆっくり回す。
※ノブは丸いまま。手はノブの周りを滑らかに移動して回転させる。指は5本で自然、変形なし。透明球の中のカプセルが慣性で少し揺れる。
ノブの回転が終わった直後、排出口（受け口）にカプセルが1つ転がって出てきて止まる（外に飛び出させない）。
キャラクターが左手でそのカプセルを自然に摘まんで持ち上げる（カプセル形状は崩さない）。
表情は控えめに満足げな微笑み、軽いまばたき1回。体の動きは最小限で自然。
【カメラ】
構図は固定、または極めて弱いスローの寄り（ズームはごく僅か）。揺れ・急ズーム・画角飛びは禁止。色味の変化なし。
【禁止】
文字・字幕・UI・追加ロゴは一切出さない。新しい物体や人物を追加しない。背景の大改変をしない。
ネガティブ：ハンドル、取っ手、クランク、レバー、棒状の持ち手、部品追加、形状変更、文字、字幕、UI、ロゴ追加、透かし、QR、バーコード、指が増える、手の崩れ、関節の破綻、顔の崩れ、ちらつき、フレームの揺れ、急ズーム、場面転換、色味の激変、ノイズ、ぼやけ

⚡ First, let's generate a video with Pollo 1.6!

First, let's create a video using the previous model, "Pollo 1.6".

From the left menu of Pollo AI, select "Image to Video". Next, for the model, choose
"Pollo 1.6".

Next, upload the "image of me, Haru, trying to turn the 'Idea Gacha Machine' while writing a note" from earlier.

Aspect ratio is set to "16:9".

Enter the common prompt from earlier into the prompt field. Turn
"Translate Prompt" ON. Set
Mode to "Pro".

I set the resolution to "720P". Then, press the
"Generate" button! The credit consumption was 50.

Pollo 1.6 generated video is here👇

The gacha nozzle didn't turn properly, and the capsule that came out flew out from the side of the main unit. Also, at the time of Pollo 1.6, there was no sound (BGM or sound effects).

🎬 Next, generating with Pollo 2.0!

Released in mid-November 2025, Pollo 2.0. Starting with this model, videos now include
sound (BGM or sound effects)!

Now, let's create the video. For the
model, select "Pollo 2.0".

The subsequent steps are exactly the same as the Pollo 1.6 from earlier. Keeping the same conditions, press
"Generate". The credit consumption this time is

40 credits.

The generated video is here!👇

In Pollo 2.0, the sound of turning the gacha was included properly.

However, the way the knob of the gacha machine turns is a bit unnatural, which is a shame...!
On the other hand, the capsule properly came out of the dispenser, so the flow of the movement itself has improved.

🎬 Finally, Pollo 2.5 video generation!

So finally, I will try generating a video using the latest model 'Pollo 2.5' that was just released on Friday, December 26, 2025!

The subsequent steps are exactly the same as the Pollo 1.6 and Pollo 2.0 ones from earlier.
Keeping the same conditions, 'Generate' !

Since it was a campaign period and credit consumption was 50% off, I was able to make it for 20 credits!

Here is the generated video!👇

The gacha knob turning that hadn't worked well before was Pollo 2.5 expressed quite naturally. Furthermore,
sound is also properly included. And most impressively,

the character developed changes in expression and emotion. The persuasiveness of the footage was clearly different.

Although the capsule became a bit huge, this seems like something that can be improved with prompt adjustment.

Overall, I felt that the quality has gone up a notch compared to Pollo 1.6 and 2.0.

✅ Summary. Pollo 2.5 is a video AI for 'mass production and growth'

This time, by comparing Pollo 1.6 → 2.0 → 2.5 using the same reference image and prompt conditions, the direction of evolution became quite clear.

⚡ Pollo 1.6 is a base model that allows for 'fast and cheap prototyping'.

However, it was prone to movement failures, such as the knob operation breaking and the capsule flying out from the side, and furthermore, it doesn't include sound (BGM/sound effects), so editing and post-processing were required to make it a finished product.

🎬 Pollo 2.0 brings this much closer to practical use. With the addition of sound, the level of completion improved, and the 'flow of movement' was improved, such as the capsule coming out of the discharge port. However, the way the knob was turned was still a bit unnatural, leaving the impression that it was just one step away...

🚀 And then Pollo 2.5. The knob turning succeeded naturally, and sound was consistently included. Even more decisive was that

changes in expression = emotions were born in the character. The persuasiveness as a video was clearly different, and it suddenly got much closer to the line of 'being able to release this as a commercial' 🔥

Of course, there is room for adjustment, such as the size of the capsules, but this is within the range that can be refined with prompts.

Overall, I feel that Pollo 2.5 has firmly evolved one step further, from an AI for creating prototypes to an
AI that can compete with posts and advertisements.

And above all, this evolution is effective here👇

🎯 Idea → Generation → Fine-tuning If you can cycle through this quickly, the number of attempts increases.

That is why Pollo 2.5 is compatible with short videos. Everyone, please mass-produce videos with

Pollo 2.5, get a hit, and let's create
that one viral video!

This article was also brought to you as a joint writing project (PR) with Pollo AI!📚