Token Ledger Digest – 2026-05-27

The most cost‑impacting change is a 50% price cut for Qwen: Qwen3.7 Max, reducing both prompt and completion costs by half.

Price Changes

• Qwen: Qwen3.7 Max – Prompt price fell from $2.50 to $1.25 per 1M tokens; completion price fell from $7.50 to $3.75 per 1M tokens. Who should care: Large‑scale Qwen users save $5.00 per 1M tokens overall.
• Qwen: Qwen3.6 35B A3B – Prompt price down 6.7% from $0.15 to $0.14 per 1M; completion unchanged at $1.00 per 1M. Who should care: Minor savings for prompt‑heavy workloads.
• Qwen: Qwen3.6 27B – Prompt price down 3.3% from $0.30 to $0.29 per 1M; completion unchanged at $3.20 per 1M. Who should care: Small reduction for prompt‑intensive tasks.
• Xiaomi: MiMo‑V2.5‑Pro – Prompt price cut 56.5% from $1.00 to $0.435 per 1M; completion price cut 71% from $3.00 to $0.87 per 1M. Who should care: Significant savings for both input and output‑heavy applications.
• Xiaomi: MiMo‑V2.5 – Prompt price down 65% from $0.40 to $0.14 per 1M; completion price down 86% from $2.00 to $0.28 per 1M. Who should care: Large cost reduction, especially for completion‑heavy workloads.
• Z.ai: GLM 4.5 Air – Prompt price down 3.8% from $0.13 to $0.125 per 1M; completion price down 1.2% from $0.85 to $0.84 per 1M. Who should care: Negligible impact; relevant only for very high‑volume users.
• DeepSeek: DeepSeek V3 – Prompt price down 28.5% from $0.32 to $0.2288 per 1M; completion price up 2.7% from $0.89 to $0.9144 per 1M. Who should care: Net saving of $0.0668 per 1M tokens; completion‑heavy users see a slight increase.

Removed Model

• Arcee AI: Trinity Large Thinking (free) – Model removed from catalog; previously offered zero‑cost access with a 262k‑token context. Who should care: Users who relied on a free large‑context model must now switch to a paid alternative.

---

Originally published at The Token Ledger. Subscribe for the daily digest.