DeepSeek V4.1 Flash 1017 Max vs Qwen3.8 Max 0902 compared

DeepSeek V4.1 Flash 1017 (Max) and Qwen3.8 Max (0902) are close rivals from DeepSeek and Alibaba. No hands-on testing went into this write-up; it summarises the recorded benchmark, price and speed data for both models. DeepSeek V4.1 Flash 1017 Max streams output at about 162 tokens per second, 4.2 times the rate of Qwen3.8 Max 0902.

Speed and latency

Once a response is under way, DeepSeek V4.1 Flash 1017 Max produces roughly 162 tokens per second against 38 tokens per second for Qwen3.8 Max 0902. That makes DeepSeek V4.1 Flash 1017 Max roughly 4.2 times as fast at writing. Before any text appears, DeepSeek V4.1 Flash 1017 Max takes 1.2 seconds and Qwen3.8 Max 0902 3.2 seconds. In an interactive app, DeepSeek V4.1 Flash 1017 Max will feel noticeably more responsive.

For an 800-token response, first token plus generation comes to roughly 6.2 seconds on DeepSeek V4.1 Flash 1017 Max and 24 seconds on Qwen3.8 Max 0902.

Qwen3.8 Max 0902 leads on the intelligence index

DeepSeek V4.1 Flash 1017 Max records 32.6 on the intelligence index and Qwen3.8 Max 0902 records 33.7. The lead is visible but narrow, so it should not decide the choice on its own. In relative terms, Qwen3.8 Max 0902 scores 3% higher.

Across the 109 active models with an intelligence score in the catalogue, DeepSeek V4.1 Flash 1017 Max ranks 45th and Qwen3.8 Max 0902 ranks 44th.

Neither model has a separate coding or math index recorded yet, so those narrower comparisons will have to wait for more data.

Which is newer

The release dates are September 2, 2026 for Qwen3.8 Max 0902 and October 17, 2026 for DeepSeek V4.1 Flash 1017 Max, about a month apart. That is close enough to count them as the same generation.

Cost: DeepSeek V4.1 Flash 1017 Max is cheaper

DeepSeek V4.1 Flash 1017 Max lists $0.30 per million input tokens and $1.20 per million output tokens; Qwen3.8 Max 0902 lists $2.00 and $6.00. Qwen3.8 Max 0902 costs 5 times as much per output token.

A reference job of 1 million input tokens and 200,000 output tokens costs $0.54 with DeepSeek V4.1 Flash 1017 Max and $3.20 with Qwen3.8 Max 0902, so DeepSeek V4.1 Flash 1017 Max saves $2.66. Scale it up to extracting fields from 20,000 pages of 1,500 tokens into 150-token records and the bill is $12.60 for DeepSeek V4.1 Flash 1017 Max against $78.00 for Qwen3.8 Max 0902.

Value for money

A blunt value measure, blended price over intelligence score, puts DeepSeek V4.1 Flash 1017 Max at $0.016 and Qwen3.8 Max 0902 at $0.089 per point. Qwen3.8 Max 0902 costs 5.5 times as much per point, which is the price of its lead.