# Baseten — DeepSeek-V4.1-Flash: more efficient prefill for coding agents

- Company: Baseten (baseten.co)
- Announced: 2026-09-11
- Category: new-model
- Coverage: not counted
- Announcement: yes
- Group: models
- Source: https://www.baseten.co/blog/deepseek-v41-flash-more-efficient-prefill-for-coding-agents
- Record: https://forck.live/items/32769-deepseek-v4-1-flash-more-efficient-prefill-for-coding-agents
- Subject: Baseten
- Open weights: yes
- Models affected: DeepSeek-V4.1-Flash, V4-Flash, V4-Pro, V4-Flash-Vision-Exp
- License: open weights
- Context window: 1M token context window

DeepSeek released the open weights for DeepSeek-V4.1-Flash, a 552B-parameter multimodal model built for coding agents, and Baseten is serving it through its Model APIs. The model uses a Causal Encoder-Decoder architecture that activates 8B parameters per token for prefill and 16B for decode, trading a heavier decode for a lighter prefill than V4-Flash, which activates 13B for both. Baseten says this suits coding agents, where agentic loops generate far more prefill tokens than decode tokens. According to DeepSeek, the global KV cache needs just a quarter of the memory of V4-Flash's, helped by Compressed Sparse Attention 2 and FP4 KV caching. On coding and agentic benchmarks, Baseten says V4.1-Flash beats V4-Pro with roughly a third of the total parameters, though a 54.8 on Automation-Bench means it fails roughly half of complex workflows. It is DeepSeek's first non-experimental model with native image input, but a ZeroBench score below 50 still warrants human review of image-based reasoning. DeepSeek has retired the V4-Flash models on its platform and routes their traffic to V4.1-Flash, with V4-Pro traffic to be rerouted beginning September 14.

## Evidence

Verbatim from https://www.baseten.co/blog/deepseek-v41-flash-more-efficient-prefill-for-coding-agents:

> DeepSeek released V4.1-Flash this week: a 552B-parameter multimodal MoE model with a Causal Encoder-Decoder architecture built for coding agents. It's available now as a Baseten Model API, with improved KV cache efficiency and asymmetric prefill/decode compute over its V4 predecessors.
> 
> This week, DeepSeek released the open weights for DeepSeek-V4.1-Flash on HuggingFace , and the model is already available on Baseten Model APIs (with Loops support coming soon). Key specifications include:
> 
> 552B total parameters
> 
> 8B active parameters for prefill, 16B for decode
> 
> 1M token context window
> 
> Multimodal input (text + image), text output
> 
> Originally popularized by Google's Gemini series, the "flash" designation describes lighter, less expensive models designed to perform similarly to larger ones. V4.1-Flash marks DeepSeek's third open-weight flash release this year, joining recent offerings from Z.ai and Alibaba as model labs deliver higher intelligence at lower price points for coding agent workloads.
> 
> Comparing DeepSeek models: old versus new
> 
> DeepSeek V4.1-Flash demonstrates improvements over V4-Flash and V4-Pro models in both text and multimodal capabilities.

---

Record: https://forck.live/items/32769-deepseek-v4-1-flash-more-efficient-prefill-for-coding-agents
Catalogue: https://forck.live/llms.txt
Current issue: https://forck.live/feed.md
