Unverified Claim Says AI Relays Dropped Luna as Low Prices Squeeze Margins

A social media post claims that some AI relay platforms have removed the Luna model because its low price makes operations unprofitable. Available evidence supports broader pricing pressure but does not verify the delisting or establish Luna’s actual costs.

Published 9/10/2026, 4:27:55 AMLast updated 9/10/2026, 4:27:55 AMEditorial publication

A social media post claims that some AI relay platforms have stopped offering the Luna model. The author argues that the model is priced so aggressively that higher usage could increase operators’ losses, with margins falling below cloud and computing costs.

No official announcement from the model provider or an identified relay platform has confirmed the alleged delisting. Other reports describe intensifying API price competition and the operating costs faced by relay services, including infrastructure, accounts, access channels and compliance. However, those reports do not establish Luna’s unit economics or prove that a particular operator is losing money.

The claim should therefore be treated as an unverified industry observation rather than a confirmed product or pricing-policy change. Confirmation would require an official pricing page, a platform notice, billing data or corroboration from independent operators.

Source evidence 6

Source review does not guarantee permanent accuracy. Consider the original source, observation time, and later corrections.

OpenAI與Anthropic大砍中階模型價格，中國AI對手翻轉成本戰局 — BigGo 財經finance.biggo.com.tw · supporting

OpenAI近日將旗下最快、最平價的模型GPT-5.6 Luna價格大幅調降80%，每百萬輸入Token降至0.20美元，每百萬輸出Token降至1.20美元。中階選項Terra的價格也調降20%。與此同時，Anthropic推出Claude Opus 5，每百萬輸入Token定價5美元、輸出Token定價25美元，僅為其最強模型Fable 5的半價。這家新創公司還取消了原定9月生效的Claude Sonnet 5漲價計畫。這波降價標誌著美國AI公司策略的顯著轉變。過去這些公司主要在效能上競爭，如今不得不正視價格因素。根據Silicon Data的Token價格指數，自7月中旬以來，美國頂尖實驗室模型的客戶實際支付價格已下跌近四分之一。Token是語言模型處理的資料單位，也是多數客戶帳單的計算基礎。競爭格局已被DeepSeek與月之暗面等中國開發商徹底顛覆。這些公司推出的開源模型能力日益強大，且可免費下載與修改。DeepSeek以價格破壞者起家，模型價格約為西方競爭對手的十分之一，採用率爆發式成長。OpenRouter的數據顯示，DeepSeek與Z.ai的中國模型在Token使用量上已超越Anthropic的Claude與OpenAI的ChatGPT，DeepSeek更成為該期間Token佔比最高的模型供應商。隨著AI帳單持續攀升，企業客戶正積極測試更便宜的替代方案。DoorDash已採用月之暗面的模型處理程式碼任務，西門子與Airbnb則在測試DeepSeek與阿里巴巴等中國供應商的模型。這股轉變恰逢OpenAI與Anthropic將部分企業客戶從定額訂閱制轉向按用量計費，企業須依據實際消耗的運算資源付費。 ### DeepSeek的策略逆轉 [...] 美國實驗室正積極運用這項論述。專注於數學、科學、程式碼與推理基準測試的Artificial Analysis發現，Anthropic的Opus 5在中努力程度下的每任務效能與成本，與月之暗面Kimi K3在最大努力程度下相當。OpenAI的GPT-5.6 Luna在最大努力程度下的表現與DeepSeek V4 Flash在最大努力程度下相近，但每任務成本略低於後者的兩倍。 AlphaSense針對246道財務分析問題的分析顯示，OpenAI的GPT-5.6 Sol比Kimi K3便宜約13%，品