Deepseek launches V4.1 Flash 1017 model with 1 M token context
Deepseek announced the V4.1 Flash model, revision 1017, a 552B‑parameter Mixture‑of‑Experts LLM that activates 16B parameters per run. The multimodal model accepts text and images and supports a 1 million token context window. Deepseek’s internal benchmarks place it between GPT‑5.6 Sol and Claude Opus 5, and it outperforms Anthropic’s LLM on task tests.
Key points
- Hardware use drops to one‑quarter HBM and one‑eighth SSD; token entry size 890 bytes
- Pricing: $0.15 input, $0.6 output per million tokens; $0.003 cache‑hit input cost
The upgrade cuts hardware needs dramatically: V4.1 uses only one‑quarter of the HBM and one‑eighth of SSD storage for its KV cache compared with the prior version, and each token entry occupies 890 bytes. Pricing is $0.15 per million input tokens and $0.6 per million output tokens, dropping to $0.003 per input token on cache hits. Search queries cost $0.75 per 1,000 queries, cheaper than Parallel’s $1. Artificial Analysis ranks DSF‑1017 #2 in the large open‑weight class with a score of 45, while Claude Opus 5.5 scores 58. The model answers correctly 43% of the time but hallucinates on 67% of non‑answers, earning 5 points on the benchmark’s main index.
Model pages: GPT-5.6 Sol → · Claude Opus 5.5 → · Kimi K3 →
The story so far
2 episodes →- Deepseek launches V4.1 Flash 1017 model with 1 M token contextthis story
Deepseek's latest AI model is smarter than ever, but its low running costs come at a price
Tom's Hardware · 23 October 2026
Loading the full article…
This text was published by Tom's Hardware and written by Bruno Ferreira. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Generative AI & Models
All →- OpenAI says its AI agents weakened internal misalignment monitoring system · 33 src
- Opinion: Jev’s hype cycle shows AI marketing overpromises and fizzles · 1 src
- OpenAI brings GPT-6 Astra, GPT-6.1 Sol, GPT-6 Luna and Codex to Google Cloud · 20 src
- Tencent releases WeVisOne-VL 4B vision-language model on Hugging Face · 1 src
- Hugging Face releases Carbon-X, a 14B parameter DNA language model · 8 src
Comments
via GitHub Discussions