Token Ledger Digest - 2026-07-01
Most Impactful Change
Anthropic Claude Sonnet Latest – Prompt price fell from $3.00 to $2.00 per 1M tokens (‑$1.00/1M); completion price fell from $15.00 to $10.00 per 1M tokens (‑$5.00/1M).
Who should care: Teams using Claude Sonnet for high‑volume generation see a 40% reduction in completion cost and a 33% cut in prompt cost.
Added Models
- Anthropic: Claude Sonnet 5 – Prompt $2.00/1M, Completion $10.00/1M, 1M‑token context.
- Google: Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image) – Prompt $0.25/1M, Completion $1.50/1M, 65 K‑token context.
Who should care: Developers needing multimodal image‑aware models (Google) or a new Sonnet variant with large context (Anthropic).
Removed Models
- Owl Alpha – Free (prompt $0.00/1M, completion $0.00/1M) removed.
- Microsoft: Phi 4 Mini Instruct – Prompt $0.08/1M, Completion $0.35/1M removed.
Who should care: Users relying on zero‑cost Owl Alpha must migrate; Phi 4 Mini users lose a low‑priced sub‑$0.50/1M option.
Price Changes (per 1M tokens)
| Model | Prompt Δ | Completion Δ |
|---|---|---|
| Z.ai: GLM 5.2 | –$0.01 | $0.00 |
| Qwen: Qwen3.6 27B | +$0.03 | +$0.02 |
| DeepSeek: DeepSeek V4 Flash | +$0.01 | +$0.02 |
| Qwen: Qwen3 235B A22B Thinking 2507 | +$0.05 | +$1.40 |
Who should care: Cost‑sensitive workloads on Qwen 235B Thinking will see completion costs rise sharply; other shifts are marginal (<$0.05/1M).
Originally published at The Token Ledger. Subscribe for the daily digest.
Top comments (0)