ZenMux is excited to introduce Ant Group inclusionAI’s Ring-1T, a trillion-parameter thinking model built on the Ling 2.0 architecture, representing the next evolution in deep reasoning and large-scale MoE (Mixture of Experts) systems. Ring-1T combines state-of-the-art reasoning capability, unprecedented training stability, and trillion-scale efficiency, setting a new benchmark for cognitive AI.
Trained with Reinforcement Learning from Verifiable Rewards (RLVR) and Human Feedback (RLHF), Ring-1T demonstrates extraordinary balance across reasoning, coding, mathematics, and natural language inference—marking a leap from the earlier Ring-flash-2.0 and Ring-mini-2.0 models.
With ZenMux’s Ring-1T API, you can expect:
- Highly Competitive Pricing: Access trillion-parameter reasoning at an unmatched cost—input at $0.56 per million tokens, output at $2.24 per million tokens.
- Massive Context Window: Up to 128K tokens, supporting long-form reasoning, codebases, and complex dialogues.
- Optimized for Deep Reasoning Tasks: Excels at mathematical competitions (IMO, HMMT), logical reasoning (ARC-AGI-1), code generation (LiveCodeBench, CodeForce), and creative writing.
Our mission has always been to make access to world-class AI simple, stable, and scalable.
Through our unified API, developers and enterprises can now seamlessly tap into Ring-1T’s trillion-parameter reasoning capabilities—without the usual headaches of infrastructure setup or model instability.
Why Ring-1T Matters
In head-to-head benchmarks, Ring-1T achieves industry-leading performance among both open-source and closed-source thinking models.
Beyond benchmark results, Ring-1T has demonstrated real-world cognitive breakthroughs:
- At IMO 2025, Ring-1T solved 4 of 6 problems in a single attempt—silver medal level—and came close to perfect proofs on geometry reasoning.
- At ICPC World Finals 2025, Ring-1T solved 5 programming problems, outperforming several leading proprietary models.
Why Use Ring-1T on ZenMux
ZenMux is built to make model access simple, stable, and developer-friendly. When you deploy Ring-1T through our platform, you gain access to a suite of features designed to streamline integration and enhance reliability.
Benefits of using Ring-1T via ZenMux:
- Unified Access Experience: Call Ring-1T like any other model—no new credentials or SDKs required. Our standardized API means you can integrate seamlessly across your existing infrastructure.
- Smart Model Routing: Automatically optimizes latency and throughput, ensuring your requests are handled efficiently and reliably.
- Output Quality Support: ZenMux’s upcoming AI model insurance ensures reliability and compensates for hallucination or latency issues.
- Multi-Model Orchestration: Combine Ring-1T with other models available on ZenMux, enabling sophisticated workflows that leverage the strengths of multiple AI systems.
In short: you focus on building—ZenMux handles the orchestration.
Get Started Immediately
- Explore: Try Ring-1T via Chat in the ZenMux Studio.
- Integrate: We recommend using the OpenAI SDK or the Anthropic SDK compatibility mode. You can also call the ZenMux API directly. For more details, please see the sample code and API documentation for Ring-1T in the ZenMux Documentation.
from openai import OpenAI
client = OpenAI(
base_url="https://zenmux.ai/api/v1",
api_key="<ZenMux_API_KEY>",
)
completion = client.chat.completions.create(
model="inclusionai/ring-1t",
messages=[
{
"role": "user",
"content": [
{
"type": "text",
"text": "What is in this image?"
}
]
}
]
)
print(completion.choices[0].message.content)Ready to experience the flow-state intelligence of Ring-1T? Start building with our API today and discover how trillion-parameter reasoning transforms your workflow.
About ZenMux
ZenMux is the world’s first enterprise-grade large model aggregation platform with an insurance payout mechanism. We provides one-stop access to the latest models across providers. When issues such as poor output quality or excessive latency occur during use, our intelligent insurance detection and compensation system automatically resolves them—eliminating enterprise concerns about AI hallucinations or instability.