- Published on
DeepSeek V4.1 Flash 1017 (Max)
- Authors
- Name
- DeepHorus
DeepSeek released V4.1 Flash 1017 (Max) on October 17, 2026. The open-weight model supports text and image input, outputs text, and has a 1-million-token context window. It scores 45 on the Artificial Analysis Intelligence Index, well above the median of 19 for comparable open-weight models.
Artificial Analysis calls the model notably fast and reasonably priced. It generates 230.4 tokens per second, far above the median of 77.9. Pricing is $0.30 per 1 million input tokens and $1.20 per 1 million output tokens, compared with medians of $0.43 and $1.68.
The model is also very verbose. On the Intelligence Index, it generated 210 million tokens versus a median of 140 million. The average cost to evaluate it on that index is $0.26 per task. Despite high token use, it remains moderately priced for its performance tier.
The release matters because it combines high intelligence, fast output, and a large context window. A 1-million-token context allows users to process long documents, codebases, or image-heavy inputs in one session. The 230.4 tokens-per-second speed could benefit real-time assistants and agentic workflows.
Artificial Analysis compares the model only with open-weight models of similar size. Its 45-point score is more than double the median of 19. The model is described as very verbose, however, which can increase output-token consumption. That trade-off may affect deployment costs in high-volume applications.
DeepSeek created the model, which was released on October 17, 2026. Its mix of speed, price, and context length could intensify competition among open-weight providers. Future evaluations will test whether verbosity hinders efficiency in production. For now, it stands out as a fast, capable, and moderately priced option.
