DEV Community

4663437Mehdi
4663437Mehdi

Posted on Originally published at 4663437mehdi.github.io

Token Ledger Digest – 2026-09-26

#ai #llm #api #news

Token Ledger Digest – 2026-09-26

Biggest cost shift: MoonshotAI: Kimi Latest – prompt price fell from $1.40 to $1.03 /1M tokens (−$0.37/M); completion price fell from $10.75 to $9.04 /1M (−$1.71/M). Teams using Kimi for long‑form generation see ~20% lower cost.

Price changes (per‑million tokens)

  • DeepSeek Pro Latest: prompt $0.389→$0.249 (−$0.140); completion $1.168→$0.748 (−$0.420); −$0.560/M
  • DeepSeek Flash Latest: prompt $0.040→$0.035 (−$0.005); completion $1.000→$0.290 (−$0.710); −$0.715/M
  • inclusionAI Ling 3.0 Flash VL: prompt $0.060→$0.021 (−$0.039); completion $0.180→$0.062 (−$0.118); −$0.157/M
  • DeepSeek V4.1 Flash: prompt $0.140→$0.300 (+$0.160); completion $0.420→$1.200 (+$0.780); +$0.940/M
  • Z.ai GLM Flash Latest: prompt $0.045→$0.040 (−$0.005); completion $0.140→$0.500 (+$0.360); +$0.355/M
  • Z.ai GLM 5.3 Flash: prompt $0.150→$0.040 (−$0.110); completion $0.500→$0.500 (0); −$0.110/M
  • Z.ai GLM Latest: prompt $0.561→$0.379 (−$0.182); completion $1.764→$1.192 (−$0.572); −$0.754/M
  • Z.ai GLM 5.3: prompt $0.840→$0.379 (−$0.461); completion $2.640→$1.192 (−$1.448); −$1.908/M
  • DeepSeek V4 Pro 0813: prompt $0.462→$0.264 (−$0.198); completion $1.386→$0.792 (−$0.594); −$0.792/M
  • DeepSeek V4 Flash Latest (~): prompt $0.030→$0.021 (−$0.009); completion $0.320→$0.320 (0); −$0.009/M
  • DeepSeek V4 Flash 0731: same as above (−$0.009/M)
  • Qwen Qwen3.6 27B: prompt $0.320→$0.320 (0); completion $2.700→$3.200 (+$0.500); +$0.500/M
  • DeepSeek V4 Pro 0423: prompt $0.936→$0.348 (−$0.588); completion $1.873→$0.696 (−$1.177); −$1.765/M
  • DeepSeek V4 Flash 0423: prompt $0.089→$0.047 (−$0.042); completion $0.177→$0.094 (−$0.083); −$0.125/M

Added models

  • TypeSafe: Jev Router – context 1M tokens; prompt $‑1.0/M, completion $‑1.0/M (placeholder).
  • Perceptron: Perceptron Mk1.5 – 36 864‑token context; prompt $0.15/M, completion $1.50/M.
  • Mistral: Devstral 2 2512 – 262 144‑token context; prompt $0.40/M, completion $2.00/M.
  • Mistral: Mistral Large 3 2512 – 262 144‑token context; prompt $0.50/M, completion $1.50/M.

Removed models

  • Nex AGI: Nex‑N2.5‑Mini (free) – 262 144‑token context, $0/M both.
  • Nex AGI: Nex‑N2.5‑Pro (free) – same.
  • Z.ai: GLM 5.2 (free) – 32 768‑token context, $0/M both.
  • Anthropic: Claude 3 Haiku – 200 000‑token context; prompt $0.25/M, completion $1.25/M.

Cheapest models today (per‑million tokens)

  1. IBM: Granite 4.0 Micro – prompt $0.017/M, completion $0.112/M
  2. OpenAI: gpt‑oss‑20b – prompt $0.018/M, completion $0.090/M
  3. Mistral: Mistral Nemo – prompt $0.019/M, completion $0.030/M

Total models tracked: 458.

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

Top comments (0)