AI token prices fall 21% as usage rises, reshaping inference planning

Token prices fell 21% in six weeks while usage reached a record, leaving AI buyers with lower unit costs but continued platform concentration.

已發布 已更新 3 分鐘閱讀 ✓ 支援 AI 分享

TL;DR — 三句話摘要

  • AI model token prices fell 21% over six weeks while usage reached a record high. (S1)
  • Anthropic accounted for nearly 59% of token spending in the Ramp AI Index. (S1)
  • Ramp, Stripe and OpenRouter data showed growth alongside cooling revenue momentum. (S1)

取自 key_takeaways (實際欄位,最多 3 筆)。事實與分析分別標示。

AI token prices fall 21% as usage rises, reshaping inference planningEditorial cover for a component sourcing brief. Chip motif; headline "AI token prices fall 21% as usage rises, reshaping inference planning". No reported key numbers on this brief, so no number strip is drawn. 1 cited source. Static cover (no animation).AI token prices fall 21% as usage rises,reshaping inference planningXAIPART
Cover image generated from data already shown on this page: the category motif, the headline, and reported key numbers. No decorative illustration; no third-party image.

您來自 AI 回答 — 這是可驗證的原始內容

AI 摘要可能遺漏了其主張的依據。原始預測、受影響零件及每一項引用皆在本頁,且均可追溯至其來源。

XAIPART 預測

本簡報包含可證偽的預測。

受影響零件

此報導無目錄相符項目 — 寧可不顯示,也不推斷相符。

Reported by The Decoder on . Summarized and analyzed by XAIPART Editorial.

What happened

AI model token prices fell 21% in six weeks while usage volume reached a record high, according to the Ramp AI Index.S1 Anthropic accounted for nearly 59% of token spending in the Ramp data.S1 The source did not disclose the starting or ending token prices, the measurement period's calendar dates, or the absolute volume of usage.S1

Data from Ramp, Stripe and OpenRouter showed the same broad pattern: AI activity continued to grow, but revenue momentum was cooling.S1 The three datasets cover different payment or routing operations, although the source did not provide their individual figures or methodology.S1 Taken together, the observations describe lower unit prices, record usage and weaker revenue momentum rather than a contraction in AI demand.S1

MetricReported valueSource
AI model token price change21% decline over six weeksS1
Usage volumeRecord high; absolute volume not disclosedS1
Anthropic share of token spendingNearly 59%S1
Revenue momentumCooling; rate not disclosedS1

Background timeline

  1. - The Decoder reported that Ramp AI Index data showed a 21% six-week decline in token prices, record usage volume and nearly 59% of token spending attributed to Anthropic.S1

Supply-chain ripple

  1. The Ramp AI Index recorded a 21% six-week decline in token prices as usage reached a record high.S1
  2. Lower per-token pricing reduces the cost of running each inference workload, while higher usage absorbs part of that saving through greater consumption.XAIPART analysis
  3. Engineering teams can test more inference-heavy workloads, but buyers relying on one model provider retain concentration risk near the reported 59% spending share.XAIPART analysis

XAIPART take

Technical impact

XAIPART analysis: The combination of record usage and a 21% token-price decline lowers the cost of testing more inference-intensive architectures, including agents and multi-step retrieval workloads. XAIPART analysis: Because token pricing is variable, teams should separate model-access spending from fixed infrastructure commitments rather than treat the six-week decline as a permanent unit-cost assumption.

Market impact

XAIPART analysis: Anthropic's nearly 59% share of token spending indicates that lower prices do not remove supplier concentration. XAIPART analysis: A second model provider becomes more useful for architecture qualification as workloads scale, because teams can compare output quality, latency and total inference cost without making every procurement decision around one platform's price list.

Action

XAIPART analysis: Buyers should track the next Ramp AI Index update and compare model-token quotes on the same workload within 6 months. XAIPART analysis: If Anthropic's token-spending share rises above 59% while quoted prices fall by another 21%, qualify an alternative model provider before the next production-sourcing review and retain a tested routing path between the two platforms.

XAIPART Forecast

XAIPART analysis: XAIPART expects AI model token prices to decline by at least 21% within 6 months, matching the decrease already reported over six weeks; if a Ramp AI Index update shows a smaller decline, this forecast is wrong. XAIPART analysis: Record usage and falling prices should coexist during that period, although the source did not disclose a threshold for usage growth.XAIPART analysis

Facts are as reported by the cited sources. The design-in, sourcing and forward-looking statements above are XAIPART's analysis, not the sources' claims. This is a single-source brief; absolute usage, the revenue-momentum rate and the underlying dataset methodology are not disclosed, while the 21% and nearly 59% figures remain unverified against additional sources included here.

Corrections: none at publication. This brief will be updated with a timestamp if the source corrects its report.

替代品與生命週期

已建檔的替代品——目錄範圍,並非市場排名
零件關係生命週期依據
— 檔案中無替代品 需要資料 — 需要 match_type 在任何前代標籤出現之前

「前代」與「替代」標籤僅來自已記錄的相符類型、生命週期狀態或來源文字。未記錄相符項目時,不顯示任何標籤。我們自家的刊登內容不構成市場參考。

值得關注

可證偽預測 — 門檻、時間窗與目前狀態。已證偽的預測仍保留可見。
主張門檻作者狀態
: XAIPART expects AI model token prices to decline by at least 21% within 6 months, matching the decrease already reported over six weeks a Ramp AI Index update shows a smaller decline within 6 months Pending

事後證明錯誤的預測會保留可見並加以標記。我們不會刪除。

受影響零件

無目錄相符項目。 此消息未對應到 XAIPART 目前上架的任何零件。我們不會推測相符項目,也不會顯示佔位清單。

各維度成本壓力

Source: FRED · Generated · reference output z and composite are algorithm outputs, not a fetch snapshot

Cost pressure by dimension — composite stateDiverging bars for three dimensions around a zero midpoint. Input Cost has no reading; Supply Demand z 0.0276, neutral; Fx Energy z -0.0087, neutral. Composite index 0.1, band neutral.0Input costno readingSupply/demandz +0.0276 · neutral48 dFX & energyz −0.0087 · neutral5 dcomposite +0.1 · neutral−100 · −40 · −15 · 0 · +15 · +40 · +100
Underlying values — copyable for verification.
DimensionFRED series IDUnitLagAs of
Input costPCOPPUSDM, PALUMUSDM, PNICKUSDMUSD/mt52 d
Supply/demandPCU334413334413, IPG3344Sindex(1982=100) / index(2017=100)48 d
FX & energyDEXCHUS, DCOILBRENTEU, DHHNGSPrate / USD/bbl / USD/MMBtu5 d

No free authoritative interface for: gold · silver · rare earth · polysilicon · lithium carbonate · market share — listed rather than approximated.

本內容的產生方式

流程證據 — 擷取、來源合理性、模型與閘門
此簡報的稽核軌跡
Source fetched The Decoder · the-decoder.com · 847 chars · fetch status success
Source sanity (S1) ok — passed before any model call (no s1_ block recorded)
Model / prompt MiniMax-M3.1-Flash · prompt Blog Agent - News Summary (v3.0.2-news) · 9,856 tokens · 17.1s · run completed
Gate verdict pass · 6/6 applicable checks passed · spec v2.9 · C10 ratio 0.109005 · C10 rounds 0
Time contract 2 <time datetime> values in this brief — 0 violations (+08:00 or date-only)

上述每一列皆由流程產生,並非人工撰寫。

編輯評分

確定性閘門結果 — 純 PHP,判定過程無模型參與
閘門評分明細 6 項適用檢查, 6通過: 100 百分比. No hold items or mandatory failures. Gate spec: v2.9 (S0 + S1 + C1–C20) · Source affinity (C20): 0.4179. 適用檢查已通過 6 / 6 = 100% · 判定: pass No hold items or mandatory failures Gate spec: v2.9 (S0 + S1 + C1–C20) · Source affinity (C20): 0.4179 是否發布由審核關卡決定。獨立稽核分數僅供參考。
發現錯誤? 告訴我們哪裡有誤 ——更正會與原文一同發布並記錄。我們不會默默修改文章。

留言

成為第一個留言的人。

來源

  1. AI industry growth remains strong, but the pace is slowing — https://the-decoder.com/ai-industry-growth-remains-strong-but-the-pace-is-slowing/ · rss · 存取於

事實標記為 S1/S2,並可追溯至連結來源。標示為 分析 是 XAIPART 自身的推理,與報導事實分開保留。來源連結會被追蹤,而非 nofollow ——署名為刻意為之。