Token Ledger Digest – 2026-06-08

Most cost‑impacting change: NousResearch Hermes 3 70B Instruct rose from $0.30 → $0.70 per M tokens for both prompt and completion (+$0.40/M). Developers using this model for high‑volume workloads should reassess cost budgets.

Price changes

Model	Change	Old price (/M)	New price (/M)	Δ (/M)	Who should care
NousResearch Hermes 3 70B Instruct	Prompt & completion ↑	$0.30	$0.70	+$0.40	Teams running large‑scale inference with this model
NVIDIA Llama 3.3 Nemotron Super 49B V1.5	Prompt ↑	$0.10	$0.40	+$0.30	Users of the Nemotron Super line
Meta Llama 3.2 11B Vision Instruct	Prompt & completion ↑	$0.245	$0.345	+$0.10	Vision‑enabled applications
Qwen Qwen3 30B A3B	Prompt ↑ $0.09 → $0.12 (+$0.03); Completion ↑ $0.45 → $0.50 (+$0.05)	—	—	—	Mid‑size LLM users
Qwen Qwen3 235B A22B Instruct 2507	Prompt ↑ $0.071 → $0.09 (+$0.019)	—	—	—	Large‑scale Qwen deployments
Qwen Qwen3.5‑9B	Prompt ↑ $0.04 → $0.10 (+$0.06)	—	—	—	Cost‑sensitive 9B‑scale users
Meta Llama 4 Scout	Prompt ↑ $0.08 → $0.10 (+$0.02)	—	—	—	Scout‑based agents
Google Gemma 3 4B	Prompt ↑ $0.04 → $0.05 (+$0.01); Completion ↑ $0.08 → $0.10 (+$0.02)	—	—	—	Gemma‑4B adopters
Google Gemma 3 12B	Prompt ↑ $0.04 → $0.05 (+$0.01); Completion ↑ $0.13 → $0.15 (+$0.02)	—	—	—	Gemma‑12B users
MoonshotAI Kimi Latest	Prompt ↓ $0.684 → $0.680 (−$0.004); Completion ↓ $3.42 → $3.41 (−$0.01)	—	—	—	Kimi users seeing slight savings
MoonshotAI Kimi K2.6	Same as Kimi Latest	—	—	—	—

Per‑token prices were multiplied by 1,000,000 for readability.

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
4. Mistral: Mistral Nemo – Prompt $0.02, Completion $0.03
5. Meta: Llama 3.2 1B Instruct – Prompt $0.027, Completion $0.201

No models were added or removed today. Total models tracked: 341.

---

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