Meituan's LongCat-2.5 lands as 1.6T MoE agent model, with DeepSeek copycat speculation
teortaxesTex · x · 2026-11-02
Meituan released LongCat-2.5, described as a 1.6T-parameter MoE (48B active) natively multimodal model built for long-horizon agents and real-world software engineering, claiming wins over Opus 5 and GPT-5.6-sol on AutomationBench and BrowserComp. Pricing: ¥2/M input tokens (¥0.04 cached), ¥8/M output.
teortaxesTex speculates Meituan largely copied DeepSeek's V4-Pro architecture and trained it on previous-generation Huawei silicon just as DeepSeek abandoned the V4 arch and moved to Ascend 950s — while remaining skeptical of the eval claims. Unverified third-party take.
More from Models
- Asked what AI firms want you to stop owning, Claude said: your own compute — alex_verem · 2026-11-02
- Is Hermes Still Worth It? Codex's $200 Tier Now Overlaps Most of Its Value — Foreign_Lead_3582 · 2026-11-02
- Compacting earlier cuts agent costs 16-23%; GLM 5.3 costs 30x less than Grok 4.7 — yb2698 · 2026-11-02
- Codex capacity errors spike whenever OpenAI hands out banked resets — CtrlAltDwayne · 2026-11-02
- Predictions: Opus 5.5-tier models at V4.1F inference cost within a year — teortaxesTex · 2026-11-02
- Google AI Pro user says Gemini keeps darkening his skin tone in generated portraits — Revolutionary-Pin664 · 2026-11-02