DeepSeek V4 Flash 0731 Scores 50 on Intelligence Index, Priced at $0.28 per 1M Tokens
- Published
- Jul 31, 2026 — 07:59 UTC
DeepSeek V4 Flash 0731, released in July 2026, has achieved an Intelligence Index score of 50 out of 101. This open weights model features a total of 284 billion parameters, with 13 billion active parameters, and outputs 210 million tokens. The model is priced at $0.28 per 1 million tokens for output, with an input cost of $0.14 per million tokens and a cache hit price of $0.003 per million tokens. The cost to evaluate the model is $72.02. This pricing is notably lower than the median output price of $1.20 per million tokens and median input price of $0.43 per million tokens observed across similar models. The median Intelligence Index score for comparable models stands at 25. DeepSeek V4 Flash 0731 is positioned as a leading model in terms of intelligence and pricing, according to an unverified claim from the source. Practitioners can leverage this model for cost-effective token usage in AI applications. For further details, see Hacker News (AI filtered).
By Callan Zhang · Jul 31, 2026 · Editorial standards →
Summarised from the primary source with AI assistance under human editorial oversight. Turing Wire is not a primary source — read the original for the authoritative account.
Source: Hacker News (AI filtered)