Token Ledger Digest – 2026-07-27

Most cost‑impacting change

• NVIDIA: Nemotron 3 Ultra – completion price dropped from $3.60 to $2.20 per 1M tokens; prompt price fell from $0.60 to $0.50 per 1M tokens. Who should care: Existing users see a ~39% reduction in generation cost; new adopters benefit from lower pricing.

Other price changes

• Z.ai: GLM 5.2 – prompt rose from $0.68 to $0.80 per 1M; completion rose from $2.15 to $2.53 per 1M. Who should care: Sizable cost increase for workloads using GLM 5.2.
• MoonshotAI: Kimi K2.7 Code – prompt fell from $0.75 to $0.73 per 1M; completion unchanged at $3.50 per 1M. Who should care: Minor prompt‑side savings.
• Qwen: Qwen3.5-35B-A3B – prompt rose from $0.14 to $0.15 per 1M; completion unchanged at $1.00 per 1M. Who should care: Negligible impact.

Removed models

• OpenAI: GPT-4o-mini Search Preview – prompt $0.15/1M, completion $0.60/1M.
• Inflection: Inflection 3 Pi – prompt $2.50/1M, completion $10.00/1M.
• Inflection: Inflection 3 Productivity – same pricing as Pi. Who should care: Anyone relying on these models must migrate to alternatives; they are no longer available for new calls.

Total models tracked: 342.

---

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