Hacker News front pageAug 14, 2026, 17:55 UTC+8

DeepSeek V4 Pro goes GA with peak/off-peak API pricing

DeepSeek peak/off-peak pricing update

DeepSeek V4 Pro is now GA, with major agent workflow gains and adjustable reasoning effort—low for simple tasks, high for daily agent work, max for complex ones. It natively supports the OpenAI Responses API and one-click Codex setup. API pricing shifts to peak/off-peak on Aug 16: off-peak is 50% cheaper. Model names stay the same; try it via Expert Mode on the app.

Why it matters: V4 Pro GA with agent hardening and a thinking-effort dial is a real feature update that matters to developers building automation on DeepSeek. Held below 85 because the post doesn't disclose GA benchmark comparisons or the actual peak/off-peak price spread — the info density i...

Read the original ↗Export Markdown

Take

The useful bit here is DeepSeek positioning V4 Pro squarely for agent workflows, not just chat. They added a reasoning effort dial across V4 Pro and V4 Flash — low for simple tasks, high for daily agent work, max for complex ones. That's more practical than a binary thinking mode toggle, especially if you're building products that mix quick lookups with multi-step agent tasks.

The native OpenAI Responses API support with one-click Codex setup is a smart move too — it cuts the migration friction for teams coming from OpenAI's ecosystem.

On pricing: peak/off-peak kicks in Aug 16, with off-peak at 50% off. V4 Pro input at $0.27/$0.14 (peak/off-peak), output at $1.07/$0.54; V4 Flash input at $0.14/$0.07, output at $0.54/$0.27. If you can batch workloads during off-peak hours, the savings are real. For real-time apps, you're still paying peak rates. I'd discount the benchmark numbers a bit — they're first-party, and we haven't seen third-party evals on specific agent tasks yet.

Source: Hacker News front page · api-docs.deepseek.com
Published: Aug 14, 2026, 17:55 UTC+8 · Aug 14, 2026