← Back to Changelog

v0.1.4 — DeepSeek V4.1 Flash 1017 for Coding

DeepSeek V4.1 Flash now runs on DeepSeek's October 17 update and becomes the coding default on Balanced and Quality, plus updated legal documents.

Changes

DeepSeek V4.1 Flash 1017

  • New version — DeepSeek V4.1 Flash now runs on DeepSeek V4.1 Flash 1017, released by DeepSeek on October 17, 2026. DeepSeek reports better results in reasoning, coding, agentic tasks and vision, with faster task completion and shorter outputs. See the benchmarks in DeepSeek V4.1 Flash 1017 Is Now on Risphai.
  • Same name and settings — The model keeps its name, and saved model choices and conversations keep working.
  • Same price — Credits per request are unchanged.

Coding Defaults

  • DeepSeek for coding — The Balanced preset now uses DeepSeek V4.1 Flash Thinking (High) for coding and the Quality preset uses DeepSeek V4.1 Flash Thinking (Max). Both used GPT-6.1 Sol before. Extreme Quality keeps GPT-6.1 Sol.
  • Lower coding estimate — The Pricing page now estimates an average coding request on Balanced at about 15 credits instead of about 30.

Reasoning Levels

  • XHigh removed — DeepSeek V4.1 Flash now offers Low, High and Max thinking. XHigh gave the same results as High, so if you used it, Risphai now uses High.

Privacy and Legal

Current models and prices are on the Information page.

releasedeepseek v4.1 flash 1017ai models