Token Ledger Digest – 2026-06-24

Most cost‑impacting change: MoonshotAI’s Kimi K2.7 Code completion price rose from $3.41 to $3.50 per M tokens (+$0.09/M). Developers using this model for long‑form generation will see a ~2.6% cost increase.

Added

• Sakana: Fugu Ultra – New model with 1M‑token context. Prompt $5.00/M, completion $30.00/M. Who should care: Teams needing ultra‑long context and willing to pay a premium for extended reasoning windows.

Removed

• EssentialAI: Rnj 1 Instruct – Prompt $0.15/M, completion $0.15/M (32k context). Who should care: Users of low‑cost, short‑context inference must migrate to alternatives.
• Prime Intellect: INTELLECT-3 – Prompt $0.20/M, completion $1.10/M (128k context). Who should care: Those relying on its mid‑range pricing for code or chat workloads.

Price Changes

Model	Change	Old ($/M)	New ($/M)	Δ ($/M)	Who should care
Z.ai: GLM 5.2	Prompt ↓	0.98	0.95	–0.03	Cost‑sensitive prompt‑heavy apps
	Completion ↓	3.08	3.00	–0.08	Completion‑focused services
MoonshotAI: Kimi K2.7 Code	Prompt ↑	0.68	0.74	+0.06	Prompt‑driven pipelines
	Completion ↑	3.41	3.50	+0.09	Long‑completion workloads
Xiaomi: MiMo‑V2.5	Prompt ↓	0.14	0.105	–0.035	Low‑cost prompt users
	Completion ↔	0.28	0.28	0.00	No impact
MiniMax: MiniMax M2.7	Prompt ↓	0.25	0.24	–0.01	Minor savings
	Completion ↓	1.00	0.96	–0.04	Completion‑heavy tasks
Microsoft: Phi 4	Prompt ↑	0.065	0.070	+0.005	Negligible effect
	Completion ↔	0.14	0.14	0.00	No impact
TheDrummer: Rocinante 12B	Prompt ↑	0.17	0.25	+0.08	Prompt‑intensive use
	Completion ↑	0.43	0.50	+0.07	Completion‑intensive use

Cheapest Models Today (per M tokens)

1. inclusionAI: Ling-2.6-flash – Prompt $0.01, Completion $0.03
2. IBM: Granite 4.0 Micro – Prompt $0.017, Completion $0.112
3. Meta: Llama 3.1 8B Instruct – Prompt $0.02, Completion $0.03

Total models tracked: 339.

---

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