DEV Community

4663437Mehdi
4663437Mehdi

Posted on • Originally published at 4663437mehdi.github.io

Token Ledger Digest – 2026-07-18

#ai #llm #api #news

Token Ledger Digest – 2026-07-18

Lead change – biggest cost impact

  • Z.ai: GLM 5.2 – Prompt price fell from $0.96/M to $0.34/M (‑$0.62/M); completion price fell from $3.01/M to $1.06/M (‑$1.95/M). Total reduction ≈ $2.57/M. Who should care: Teams running high‑volume inference on GLM 5.2 see per‑million‑token costs drop by over 85%; ideal for batch‑processing or chat‑bot workloads.

Added models

  • Thinking Machines: Inkling – Prompt $1.00/M, completion $4.05/M, 1M‑token context. Who should care: Users needing very long context (1 M tokens) with moderate pricing.
  • OpenRouter: Auto Router (Beta) – Prompt ‑$1.00/M, completion ‑$1.00/M (negative pricing indicates a credit). Context 2M tokens. Who should care: Experimental users who can earn credits while testing routing logic.

Removed models

  • NVIDIA: Llama 3.3 Nemotron Super 49B V1.5 – Was $0.40/M prompt & completion, 128K context. Who should care: Migrate to alternatives if you relied on this specific Nemotron variant.
  • Meta: Llama 3.2 11B Vision Instruct – Was $0.345/M prompt & completion, 128K context. Who should care: Vision‑enabled applications must switch to another vision model or adjust budgets.

Other price changes (prompt/completion, $/M)

  • MoonshotAI: Kimi K2.7 Code – ↑ Prompt $0.75 → $1.00 (+$0.25); ↑ Completion $3.50 → $4.40 (+$0.90).
  • Google: Gemma 4 31B – ↓ Prompt $0.22 → $0.12 (‑$0.10); ↓ Completion $0.55 → $0.37 (‑$0.18).
  • Qwen: Qwen3.5‑27B – ↑ Prompt $0.20 → $0.26 (+$0.07); ↑ Completion $1.56 → $2.60 (+$1.04).
  • Z.ai: GLM 4.7 Flash – ↑ Prompt $0.060 → $0.0605 (+$0.0005); completion unchanged $0.40.
  • Qwen: Qwen3 30B A3B – ↑ Prompt $0.12 → $0.13 (+$0.01); ↑ Completion $0.50 → $0.52 (+$0.02).
  • Qwen: Qwen3 14B – ↑ Prompt $0.10 → $0.23 (+$0.13); ↑ Completion $0.24 → $0.91 (+$0.67).
  • Mancer: Weaver (alpha) – ↓ Prompt $0.75 → $0.50 (‑$0.25); ↓ Completion $1.00 → $0.75 (‑$0.25).

Who should care: Developers using any of the above models should update cost estimates; increases may affect budget‑tight projects, decreases offer savings for scale‑out workloads.

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

Top comments (0)