The Token Ledger Digest – 2026-07-03
Biggest cost impact: Z.ai GLM 5.1 saw a sharp price drop.
- Prompt: $0.975 → $0.966 / 1M tokens (−$0.009)
- Completion: $4.30 → $3.036 / 1M tokens (−$1.264) Who should care: Teams running long‑form generation or chat workloads on GLM 5.1 will see per‑million‑token costs fall by ~30%.
Other price changes
| Model | Prompt old/new ($/1M) | Completion old/new ($/1M) | Net Δ ($/1M) | Who should care |
|---|---|---|---|---|
| MoonshotAI Kimi Latest | 0.55 → 0.66 (+0.11) | 3.20 → 3.41 (+0.21) | +0.32 | Users of Kimi Latest facing higher per‑token spend. |
| MoonshotAI Kimi K2.6 | 0.55 → 0.66 (+0.11) | 3.20 → 3.41 (+0.21) | +0.32 | Same impact as Kimi Latest. |
| DeepSeek V4 Flash | 0.089 → 0.090 (+0.0001) | 0.18 → 0.18 (0) | +0.0001 | Negligible cost rise; relevant for high‑volume calls. |
| Qwen Qwen3 VL 8B Instruct | 0.08 → 0.117 (+0.037) | 0.50 → 0.455 (−0.045) | −0.008 | Slight overall saving; vision‑language workloads benefit. |
Added models
- Poolside: Laguna XS 2.1 (free) – Prompt $0.00, Completion $0.00 / 1M tokens. Ideal for zero‑cost prototyping.
- Poolside: Laguna XS 2.1 – Prompt $0.06, Completion $0.12 / 1M tokens. Low‑cost option for latency‑sensitive apps.
Cheapest models today (per‑million‑token total)
- inclusionAI: Ling‑2.6‑flash – $0.04
- IBM: Granite 4.0 Micro – $0.129
- Meta: Llama 3.1 8B Instruct – $0.05
Total models tracked: 340. No removals reported.
Originally published at The Token Ledger. Subscribe for the daily digest.
Top comments (0)