In today’s AI landscape, most platforms still compete on isolated improvements – faster inference, better benchmarks, incremental gains.
Agnes AI is a Singapore-based company that has developed and trained its own multimodal AI models in-house, positioning itself as a locally built player in the global AI ecosystem.
Agnes AI is taking a different approach.
With its model suite now available on Zenmux through API token access, Agnes AI is expanding developer access to its proprietary models while separately launching Agnes Token Plans through its official website.
Together, these moves reflect a broader strategy: making full-stack AI more accessible, predictable, and practical for real-world usage.
From Models to Systems
The shift Agnes is betting on is simple: developers no longer need better models in isolation – they need systems that work in practice.
Most AI workflows today are fragmented. One tool for reasoning, another for generation, another for execution. Integration becomes the real bottleneck.
Agnes addresses this by packaging its capabilities into a single environment where reasoning, generation, and action are tightly connected.
The result is less about model selection and more about workflow execution.
This approach reflects the company’s core principles – AI neutrality, parity, and inclusion – now applied at the infrastructure level.
Instead of limiting access through pricing or ecosystem lock-in, the platform is designed to be accessible, cost-efficient, and globally deployable.
From API Access to Subscription-Based AI Usage
Through Zenmux, developers can access Agnes’s model suite using API token access, enabling integration into external products, workflows, and developer environments.
Separately, Agnes has launched Agnes Token Plans through its official website. Starting at US$4 per month, the subscription structure is designed for high-frequency, lightweight workloads, while also supporting more advanced professional use cases.
This pricing model reflects a strong focus on cost efficiency, making multimodal AI more accessible without the cost barriers that have historically limited adoption across similar platforms.
The key shift is not only economic, but practical: AI becomes something users can integrate into ongoing workflows with more predictable costs.
The platform supports high-throughput processing, access across Agnes’s text, image, and video models, OpenAI API compatibility, and scalable usage from everyday productivity to more complex production workflows.
The Zenmux Launch and Early Developer Traction
The Zenmux launch includes access to Agnes’s agent-based models and multimodal generation capabilities, forming a pipeline across text, image, and video.
What stands out is not just the coverage, but the adoption signal. Within the first week, API usage surged beyond several comparable model providers, suggesting that demand is shifting toward integrated systems rather than standalone models.
This aligns with a broader market movement: developers are optimising less for theoretical performance and more for how quickly systems can be deployed and used.
New Applications Powered by Proprietary Models: Pavo and Echo
Beyond infrastructure, Agnes is extending its capabilities into the application layer with two new products: Echo and Pavo, both powered by its proprietary multimodal models.
Echo is positioned as an immersive AI character interaction platform, where users can create digital personas and engage in narrative-driven experiences. The system generates synchronised video and voice in real time, enabling interactive storytelling and personalised, cinematic interactions.
Pavo focuses on AI-powered image and video creation for creators and businesses. Built on Agnes’s proprietary models, it enables fast generation of cinematic content from simple inputs, supporting workflows such as marketing asset creation, social media production, and rapid prototyping.
Together, these applications demonstrate how Agnes’s model ecosystem extends beyond infrastructure into real user-facing experiences.
A Shift Already Underway
The Zenmux release and the official Token Plans launch reflect a structural shift in AI adoption.
As model capabilities converge, differentiation is moving toward usability, integration speed, and cost efficiency.
Agnes positions itself directly in this transition.
Instead of competing model-by-model, it is building an environment where AI becomes part of a working system – one that developers can deploy, extend, and scale without rebuilding infrastructure around it.
At the same time, the company has reached approximately US$20 million in annual recurring revenue (ARR), signaling commercial traction as it expands its global developer and application ecosystem.
Developed, trained, and deployed in Singapore, Agnes also reflects the growing capability of locally built AI technologies to compete on a global stage.
Featured image credit: Edited by Fintech News Singapore, based on image by Agnes
