Meituan's LongCat-2.0 Stealth AI Tops OpenRouter with Bargain Pricing
Meituan unveiled LongCat-2.0, a 1.6 trillion-parameter AI model that ran anonymously as Owl Alpha, topping OpenRouter leaderboards. It undercuts competitors like GPT-5.5 with API pricing at $0.75/$2.95 per million tokens, though our coding test revealed some logical flaws.
Quick Take
Meituan's LongCat-2.0 secretly ranked first on Hermes Agent before reveal.
1.6T parameter MoE model trains on Chinese ASICs, activations 33–56B per token.
API starts at $0.75/$2.95 per million input/output tokens, far below GPT-5.5.
Coding test shows cost-effective mid‑tier quality, but with frustrating logic errors.
Market Impact Analysis
NeutralArticle focuses on a generic AI model with no direct crypto tie; negligible impact on digital asset markets.
Speculation Analysis
Key Takeaways
- Meituan’s LongCat-2.0 topped OpenRouter’s Hermes Agent leaderboard during a two-month anonymous run as Owl Alpha.
- The 1.6-trillion-parameter mixture-of-experts model activates just ~48B parameters per token, slashing compute costs.
- API pricing starts at $0.75/$2.95 per million input/output tokens, massively undercutting GPT-5.5 and Claude Sonnet 5.
- Coding test finds solid mid-tier performance with some logical flaws, but the value is compelling.
- Trained on domestic Chinese ASICs, the model signals growing independence from Nvidia hardware.
What Happened
Meituan revealed LongCat-2.0 on June 30, ending speculation about the anonymous “Owl Alpha” model that had dominated OpenRouter rankings. The 1.6-trillion-parameter MoE system quietly accumulated real-world usage for two months before the official launch. It’s now available via API, directly challenging top-tier models from OpenAI and Anthropic. The stealth strategy allowed Meituan to benchmark and refine the model under live conditions.
The Numbers
LongCat-2.0’s architecture is massive but efficient: only about 48 billion of its 1.6 trillion parameters activate per token, keeping inference costs low. Pretraining used over 35 trillion tokens on a cluster of 50,000+ domestically produced accelerators. The standard API price is $0.75 per million input tokens and $2.95 per million output — a fraction of GPT-5.5’s $5/$30. On OpenRouter, it ranked first on Hermes Agent, second on Claude Code, and third on OpenClaw by monthly call volume.
Why It Happened
Meituan, a giant in food delivery, has been expanding in AI. The stealth launch under the Owl Alpha alias allowed the model to build a reputation organically before facing direct comparisons. The use of Chinese ASICs for both training and inference marks a shift: unlike rivals that still depend on Nvidia hardware, LongCat-2.0’s end-to-end domestic stack demonstrates progress in China’s chip ecosystem. Aggressive pricing is a clear market penetration tactic, undercutting Western incumbents.
Broader Impact
LongCat-2.0’s debut pressures AI pricing downward and validates Chinese ASIC capabilities. It could accelerate adoption among budget-conscious developers, especially in Asia. However, coding tests reveal gaps in complex reasoning, suggesting it’s currently a mid-tier alternative for tasks not requiring elite performance.
What to Watch Next
- Pricing wars: Will OpenAI and Anthropic respond with cuts or keep their premium positioning?
- ASIC viability: The model’s stability on domestic hardware could embolden more Chinese firms to build without Nvidia.
- Performance parity: Continued benchmarking will show if LongCat-2.0 closes the gap with top models in coding and logic.
This article is for informational purposes only and does not constitute financial advice.
Always late to trends?
Join for the latest news, insights & more.
Disclaimer: Bytewit is an independent media outlet that delivers news, research, and data.
© 2026 Bytewit. All Rights Reserved. This article is for informational purposes only.
