Agnes AI Free API Test - Text/Image/Video Models All Free and Available

Two days ago, I saw a message in the early morning that instantly woke me up.

A company ranked 9th in the worldAI LabQuietly put all three of his core modelsfreeAlright.

Text, images, videos, all of themfreeI'm not kidding you, it's not a stripped-down or limited version, you can use it as soon as you register.

What's going on here? This style isn't familiar; it's definitely not from China. This team from Singapore is quite formidable; they've managed to get all three models working together and still garner international recognition.

There has been a lot of discussion about it online recently.

In today's article, K-sister will conduct a real-world test to see if it's true.freeYes, it really works.

Longtime readers know that K-sister reads...AII have a habit: I never believe official announcements, and no matter how flashy a demo is, I don't get excited. I only pay attention to two things: chart rankings and blind review scores.

The three lists that Agnes entered this time all happen to measure "work ability".

Let's first talk about the text model Agnes-2.0-Flash, which is ranked Claw-Eval.

The Claw-Eval leaderboard is widely recognized in the developer community.AgentPractical test report. Agnes-2.0-Flash has implemented certain features in some tasks.Gemini Both Flash and MiniMax M2.7 were overclocked.freeThe fact that this model can compete head-to-head with the top-tier paid models on this list speaks volumes about its quality.

Next, let's look at the image model Agnes-Image-2.0-Flash. It's on the Artificial Analysis image editing blind review list.

The rule is that no one is allowed to label the generated images as being from which company created them; the scorers rely solely on visual judgment. Brand prestige and existing biases completely fail here.

Agnes managed to squeeze into the top ranks of this list. Think about it, the image editing field is full of established players who only work with images. For a company that also handles text and video to suddenly appear and claim a top spot seems a bit unethical.

Finally, the video model Agnes-Video-2.0 was used in Artificial Analysis.Image to Video(With audio) Leaderboard.

Creating videos from images isn't difficult, but creating videos with integrated audio and synchronized sound is a different story. You can throw in static images and get dynamic short videos with natural movement and ambient sound effects. For teams that need to produce materials in bulk, this capability is invaluable.freeThe API offers unbeatable value for money.

The text, image, and video charts each appear on the list, corroborating each other. A single subject's success can be attributed to luck, but achieving top marks in all three subjects is solely a testament to skill—there's nothing more to say about that..

After looking at the list, K-sister got itchy to try it herself. Talk is cheap, so I picked some of the most challenging scenarios for each of the three models and tested them out for everyone.

Agnes 2.0 text model in Flash

AIMarketing overseas tools website

NowAIOverseas expansion tools are very popular, so I created a website with 3D effects. Visually, the particle-like Earth rotates with connecting lines, giving it a strong futuristic feel. Moreover, it's not a dry, static webpage; interactive buttons are clickable, pop-ups have entrance animations, numbers jump and increment, and cards have light effects when hovered. The structure is also very complete, guiding users step-by-step from the homepage, functions, workflow, pricing to CTA.

AIMarketing Content Generator

The highlight of this case is that it...AIThe concept of content generation has been made into a real interactive experience. You input product information, select product category and style, and click Generate to see the process of copywriting being generated.

The most obvious example is those 3D flip cards; you see the product on the front, and when you flip them over...AIThe copy I wrote makes this interaction very intuitive.

AgentProduct research and analysis report

This case transformed an in-depth study of 17 products into a beautifully presented interactive presentation. I handled everything from searching, organizing, and analyzing to outputting the complete HTML slides, without relying on any external tools.

The report progresses logically and systematically, from market overview to product analysis, ranking comparison, pricing strategy, capability matrix, trend insights, and overseas opportunities. Agnes 2.0 Flash's structured presentation is quite impressive.

Image model Agnes image 2.1 Flash

Miniature World × Fantasy Architecture

The case chosen is "Glass Castle on a Dewdrop," a concept that pushes the limits of material complexity that models can achieve.

Friends, look, there must be at least three transparent materials interacting at the same time in the picture. The dew is the refraction of water, the castle is the light transmission and reflection of glass, and the crystal tower also has internal light.

When these three elements are placed together, the light passes through the dewdrop, refracts onto the castle, and then reflects back to the surroundings, creating a physically complex situation. This tests the model's depth of understanding of the materials in the real physical world.

Cyber neon city night view

Suitable for use as header images and video covers for technology-related content, it has a strong visual impact and can immediately demonstrate the model's skill.

One image simultaneously addresses three of the most challenging aspects. Neon lights are multi-source, and the model must ensure that the light from each sign correctly hits the wall, umbrella, and windbreaker, while the light colors blend together; otherwise, the entire design will fall apart.

The most ruthless are the Chinese characters on neon signs; the curved, glowing font is most prone to miswriting strokes or becoming blurred into light spots, making it difficult to distinguish between "The Future Has Arrived" and other similar characters.AI The word "city" is clearly written and even has a glowing quality; the text rendering is basically top-notch.

Amazon e-commerce main image

Amazon's main image is the standard test for business images. The rules are strict, the margin for error is low, and whether it performs well or not is immediately apparent.

Prompt wordsPure white background, a dark gray wireless earphone charging case, the lid is open, the product is centered, commercial studio shot, high-definition details.

The purity of the white background is a testament to skill. The ability to render the background as a true 255 pure white while maintaining sharp, clear edges on the product demonstrates a keen understanding of the commercial context of "studio shooting." The matte finish of the headphone case and the internal metal contacts should exhibit different reflective qualities. Models that can differentiate their textures demonstrate a top-tier understanding of materials.

This set of models passed the test cleanly and efficiently, demonstrating that the model can withstand the real demands of e-commerce sales.

Agnes Video v2.0 video model

First, create a 5-second short video with the camera slowly panning from right to left, the content moving naturally, accompanied by ambient sound effects from Tokyo's nighttime city streets.

The camera pans are smooth, without any screen tearing or frame skipping. (Paired with...)freeWith the API's buff, the cost of mass production can be ignored.

Let's look at a more cinematic shot: the overall atmosphere is very grand. Aliens land on Mars in a spaceship, they emerge from the spaceship, the protagonist walks slowly forward, and yellow sand swirls in the wind, creating a very blockbuster feel.

The following case tests a person's expressiveness: A young woman stands by the window, smiling as she turns to look at the camera, which slowly zooms in. With tears in her eyes, she says, "I'll take care of you."

The visual style is reminiscent of 1970s American films, depicting a couple dancing the tango to music at a ball, while others watch and applaud. Everyone's expressions, movements, and demeanor are perfectly integrated into the scene.

Finally, let's look at the Miyazaki-style animation effects in the film: a little girl squats down in front of a stray dog and says, "I'll take you home." The dog whines and nods.

The most brilliant aspect of this case study is its breadth of coverage, ranging from zero-cost establishing footage to sci-fi blockbusters, from emotional performances to retro ensemble pieces, and even animation styles. It perfectly validates each of the five most valuable uses of video models.

Whether it's selling products, creating short dramas, popularizing science, or creating stylized content, they can handle it all.

onefreeThe fact that the video model can produce this kind of effect speaks volumes about its quality.

After discussing the rankings and running the case, K-sister has some thoughts to share.

In the past year or so, the model's capabilities have increased dramatically, and so have the API prices.AgentIt has become the mainstream approach, but it's inherently a major consumer of tokens, involving multiple steps, multiple tools, and repeated calls, each time burning through real money. When K-sister runs complex tasks using the paid model, she really needs to mentally prepare herself before clicking "run."

If this continues, it can be played on a large scale.AgentIn the end, only a few large companies with ample budgets will remain. Small and medium-sized developers, independent creators, and students are almost being deprived of even the right to try and fail.

Agnes this timefreeOpenness—what K-sister sees is not marketing, but an attitude:AIThis card game shouldn't be limited to those who can afford the chips.

A small team of two, after integrating Agnes, is on the same starting line as a large company with hundreds of employees in terms of modeling capabilities. Ideas in your mind can be tested and validated immediately, and the word "trial and error cost" can be temporarily deleted from your dictionary.

K-sister herself has already switched to the default model for her daily experiments. The sense of freedom from having to stare at the bills and breathe is just amazing.

Agnes three modelsfreeThe API is now fully open, so you can head over to the official website to take advantage of the deals!

Original link:This company is ranked 9th globally. AI Lab, API Completefree…