DeepSeek Model Names to Be Deprecated in 2026
teortaxesTex · x · 2026-07-18
According to the API documentation in the screenshot, deepseek-chat and deepseek-reasoner will be deprecated on 2026/07/24 15:59 UTC, corresponding to the non-thinking and thinking modes of deepseek-v4-flash, respectively.
The screenshot also lists the differences in capabilities and pricing between the old and new versions:
- FIM Completion only supports non-thinking mode
- 1M input tokens: cache hit / miss prices are $0.0028 / $0.14 (the other column shows $0.003625 / $0.435)
- 1M output tokens: $0.28 / $0.87
- Concurrency limit: 2500 / 500
The core takeaway from this post is that DeepSeek is pushing forward with the migration of its API model names and interface layer.
More from Infra
- AI Infra Meme: While Countries Need TW-Scale Power for Datacenters, People Argue About Lightbulbs — mgostIH · 2026-10-27
- Legora runs 2,000 parallel agent evals on Modal Sandboxes for legal AI — blelbach · 2026-10-27
- Cloud release velocity scores: AWS 95, Azure 90, Google Cloud 82 — DavidLinthicum · 2026-10-27
- Selling a 5090 to scale local inference: one business weighs 5 hardware paths — Anarchaotic · 2026-10-27
- AWS tutorial: fleet-wide edge CV model lifecycle with SageMaker SDK v3, ~$0.07 per run — AWS ML Blog · 2026-10-27
- Qwen3.8-27B hits 40 tok/s on Jetson Thor with INT4 + MTP, accuracy beats FP16 — No-Dragonfly6246 · 2026-10-27