The Token Ledger Digest – 2026-06-10

Biggest cost impact: MiniMax: MiniMax M2.5 – completion price fell from 1.15e-06 to 9e-07 ($1.15 → $0.90 per 1M tokens, –$0.25/M). Prompt unchanged at $0.15/M. Who should care: Teams running high‑volume completion workloads on M2.5 will see ~22% lower per‑token cost.

Other price changes

• MiniMax: MiniMax M2.7 – prompt $0.279 → $0.270/M (‑$0.009/M); completion $1.20 → $1.08/M (‑$0.12/M).
• MoonshotAI: Kimi K2.5 – prompt $0.40 → $0.35/M (‑$0.05/M); completion $1.90 → $1.89/M (‑$0.01/M). Who should care: Cost‑sensitive applications using these models can trim expenses, especially for completion‑heavy tasks on M2.7.

Added models

• Anthropic: Claude Fable Latest – prompt $10.00/M, completion $50.00/M.
• Anthropic: Claude Fable 5 – same pricing. Who should care: Users needing 1M‑token context with premium pricing; suitable for long‑form generation where cost is secondary to capability.

Removed models

• Z.ai: GLM 5V Turbo – prompt $1.20/M, completion $4.00/M.
• Z.ai: GLM 4.5 Air (free) – $0/M both.
• Z.ai: GLM 4 32B – prompt $0.10/M, completion $0.10/M.
• Arcee AI: Maestro Reasoning – prompt $0.90/M, completion $3.30/M. Who should care: Anyone relying on these low‑cost or free options must migrate to alternatives; the free model’s removal may increase expenses for budget‑conscious projects.

Total models tracked: 339.

---

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