OrcaRouter's Model Leaderboard: GPT-5.6 Luna Stays on Top Amidst Fierce Competition

OrcaRouter's Ongoing Leadership in AI Model Rankings

In the fast-evolving landscape of artificial intelligence, the OrcaRouter Model Leaderboard has recently reinforced its position as a trustworthy benchmark for assessing the capabilities of AI models. According to FlashLabs, the exclusive distributor of the OrcaRouter in Japan, the state-of-the-art AI model GPT-5.6 Luna, developed by OpenAI, has successfully maintained its overall top ranking with a remarkable score of 75.2 points as of September 1, 2026.

Consistent Excellence Amidst Competition

As major advancements in AI technology surged in late August 2026, new models such as Claude Opus 5 by Anthropic and Kimi K3 by Moonshot AI were unveiled, creating a highly competitive environment. Despite this influx of new contenders, GPT-5.6 Luna's impressive performance enabled it to fend off challenges from Claude Opus 5, which holds the second position with a score of 72.4 points, and GPT-5.6 Sol in third place with 71.8 points. Such resilience illustrates not only the model's technical prowess but also its practical application in real-world scenarios.

Understanding the Composite Index

The cornerstone of the OrcaRouter Model Leaderboard is its unique Composite Index, which evaluates AI models based on a multifaceted approach rather than relying solely on laboratory benchmarks. This index is a balanced integration of four key components:

1. Human Preference (40%): Evaluated through blind tests and community voting reflects user preferences and satisfaction.
2. Independent Benchmarks (30%): Scores from third-party assessments, like the Artificial Analysis Intelligence Index.
3. Real-World Usage Data (20%): Metrics grounded in practical traffic data from the OrcaRouter platform.
4. Ecosystem Penetration (10%): Based on metrics such as OpenRouter token share and the model’s actual adoption rate.

This comprehensive scoring method enhances transparency and fosters a better understanding of which models excel in practical use, rather than just in controlled conditions.

The Current Landscape of AI Models

With the recent developments in the AI sector, the rankings are reflecting a dynamic environment. The top ten models as of September 1, 2026, according to the Composite Index, are as follows:

Rank	Model	Composite Index
--	-----	------
1	GPT-5.6 Luna (OpenAI)	75.2
2	Claude Opus 5 (Anthropic)	72.4
3	GPT-5.6 Sol (OpenAI)	71.8
4	GPT-5.4 Pro (OpenAI)	71.4
5	Kimi K3 (Moonshot AI)	70.8
6	GLM-5.3 (Z.ai)	70.3
7	GLM-5.3 Flash (Z.ai)	69.8
8	GLM-5.2 (Z.ai)	69.3
9	Qwen3.7 Max (Alibaba)	69.3
10	Grok 4.6 (xAI)	67.3

Future Directions for OrcaRouter

Looking ahead, the OrcaRouter Model Leaderboard is set to continue its mission of promoting transparency in model selection. By ensuring that users are not misled by singular benchmarks, OrcaRouter helps facilitate access to over 200 AI models through a single gateway. This combined platform lays a strong foundation for businesses seeking innovative solutions tailored to their needs.

About OrcaRouter

The OrcaRouter is an AI inference gateway developed by the American research institution Continuum AI, with FlashLabs being its exclusive distributor for the Japanese market. It features an API compatible with OpenAI, enabling access to over 200 models including notable names like Claude Opus, Google Gemini, and many more. FlashLabs continuously updates the model catalog shortly after new releases, keeping users informed of the latest advancements in AI applications.