The Singapore-born multimodal AI lab has redefined competitive AI pricing, ranking among the world’s Top 10 AI labs at half the cost of former frontrunner DeepSeek.
— Agnes AI, a Singapore-based AI model company that develops and trains its multimodal foundation models in-house, today revealed top 10 global placements on two leading independent AI benchmarks: Claw-Eval and Artificial Analysis. The standings place Agnes AI alongside Anthropic, OpenAI, Google, and DeepSeek — a landmark moment for Singapore’s position in frontier AI.
The results build on Agnes AI’s prior top 10 finish on PinchBench, the leading benchmark for agentic AI. Three consecutive top 10 entries across three independent benchmarks confirm Agnes AI as a sustained presence at the frontier of global AI.
About the Benchmarks
Claw-Eval and Artificial Analysis are open third-party rankings used by researchers, engineers, and institutions worldwide to gauge real model capability. A top 10 finish on either signals frontier performance — achieving it on both, across distinct model categories, is exceptionally rare. Claw-Eval benchmarks agentic systems underpinning autonomous agents such as OpenClaw and Hermes, assessing reliability and execution on multi-step real-world tasks. Artificial Analysis rates text and image models across global labs; its image evaluations use blind review by hundreds of independent assessors.
Benchmark Rankings
Claw-Eval — Agnes-2.0-Flash (Text / Agentic Model)
• Agnes-2.0-Flash has achieved a global top 10 position on Claw-Eval, a benchmark that assesses reliability, task execution, and real-world agentic performance across multi-step workflows.
• Agnes AI holds a spot in the global top 10 — positioned alongside Anthropic, OpenAI, Xiaomi, and DeepSeek, and ranked above Gemini and MiniMax.
• Reference:https://claw-eval.github.io/
Artificial Analysis — Agnes-Image-2.0-Flash (Image Model)
- Claw-Eval — Agnes-2.0-Flash (Text / Agentic Model): Global top 10, positioned alongside Anthropic, OpenAI, Xiaomi, and DeepSeek, and ranked above Gemini and MiniMax.
Reference: https://claw-eval.github.io/
- Artificial Analysis — Agnes-Image-2.0-Flash (Image Model): Global top 10 for image editing, rated through independent blind review. Agnes-Image-2.0-Flash is a distinct model from Agnes-2.0-Flash; together they cover Agnes AI’s full multimodal capabilities across text/agentic and image domains.
Reference: https://artificialanalysis.ai/image/leaderboard/editing
Singapore’s First AI Lab to Enter the Global Top 10
Agnes AI is the first Singapore-founded AI lab to reach the global top 10 on international AI benchmarks — and has now done so three times running. The company built its entire multimodal stack — text, image, and video — from the ground up in Singapore. These results confirm a lab conceived, developed, and trained in Singapore belongs on the same leaderboard as the world’s most established AI companies.
“Singapore’s National AI Strategy 2.0 set a clear ambition: for Singapore to be a creator of AI, not simply a consumer of it. These results show that ambition becoming reality. Agnes AI was conceived here, built here, and is now counted among the world’s top 10 AI labs. Our mission is to develop AI that is globally competitive, openly available, and free from dependence on any single power or ecosystem. This is only the beginning.” — Bruce Yang, CEO, Agnes AI
Industry-Leading Pricing: Top 10 Performance, Fraction of the Cost
Both models are top 10 globally and among the most affordably priced in their classes. Full pricing: https://platform.agnes-ai.com/settings/profile
Agnes-2.0-Flash — Text Model API Pricing
Input: US$0.03 per million tokens. Output: US$0.15 per million tokens. Less than half the cost of DeepSeek-V4 Flash (US$0.28/M output) — and a fraction of comparable Anthropic and Google models. For teams running high-volume agentic workloads, that gap translates directly to the bottom line.
Agnes-Image-2.0-Flash — Image Model API Pricing
US$3.00 per 1,000 images, against a market average of roughly US$30.00 — over 70% cheaper. The contrast is sharper at the top: Google’s Nano Banana at US$39.00, ByteDance Seed’s Seedream 4.5 at US$40.00, OpenAI’s GPT Image 2 at high quality at US$211.00 — over 70 times the price of Agnes-Image-2.0-Flash. Not a marginal saving — a fundamental reset of what world-class image AI should cost.
This reflects a founding principle at Agnes AI: world-class AI performance should not come with a premium price tag. Both models deliver frontier-grade capability at a fraction of what dominant platforms charge.
Developed, trained, and deployed in Singapore, Agnes AI demonstrates that frontier AI need not be costly or imported. Made in Singapore. Recognised worldwide.
About Agnes AI
Agnes AI is a Singapore-based AI model developer building proprietary multimodal foundation models and supporting infrastructure for high-efficiency AI across consumer and enterprise applications. Its portfolio spans text, image, and video generation. Driven by a mission of AI inclusion, parity, and neutrality, Agnes AI creates scalable AI products for individuals, developers, and organisations worldwide.
Contact Info:
Name: Victoria, Marketing
Email: Send Email
Organization: Agnes AI
Website: https://agnes-ai.com/
Disclaimer:
This press release is for informational purposes only. Information verification has been done to the best of our ability. Still, due to the speculative nature of the blockchain (cryptocurrency, NFT, mining, etc.) sector as a whole, complete accuracy cannot always be guaranteed.
You are advised to conduct your own research and exercise caution. Investments in these fields are inherently risky and should be approached with due diligence.
Release ID: 89192496
If you detect any issues, problems, or errors in this press release content, kindly contact error@releasecontact.com to notify us (it is important to note that this email is the authorized channel for such matters, sending multiple emails to multiple addresses does not necessarily help expedite your request). We will respond and rectify the situation in the next 8 hours.
