Deepseek V4.1 Flash

TL;DR: DeepSeek Harness in 30 Seconds DeepSeek Harness (dsh) is DeepSeek’s open-source AI agent: it writes and tests code, works with documents and spreadsheets, researches with citations and runs scheduled tasks. MIT licensed. The app is free, the model is not. It runs on a DeepSeek API key. The default model, V4.1 Flash, costs $0.15 input / $0.60 output per million tokens off-peak. On September 29, 2026 DeepSeek shipped a desktop app for Windows and macOS (Apple silicon) in worldwide public preview. No Node.js needed. No Claude or ChatGPT subscription login. You can add Anthropic, OpenAI, Kimi and GLM, but only with API keys. Everything is a plugin, and you can have the agent write a plugin for you through chat (Creator mode). It is still a developer preview: breaking changes are expected and DeepSeek says it has not been security audited. An open-source Claude Code that you point at whatever model you like, paying only for tokens. That is the pitch behind DeepSeek Harness, and it explains why the repository is closing in on 240,000 GitHub stars about seven weeks after it went public. Until today you needed Node.js and a terminal to run it. The new desktop app removes that step. Below: what it costs in practice, how to set it up in about five minutes, and when it makes more sense than Claude Code. ...

Summary: DeepSeek V4.1 Flash in 30 Seconds DeepSeek-V4.1-Flash shipped on September 10, 2026. License is MIT, weights are on Hugging Face. 552B backbone parameters, but only 8B active while reading input and 16B while generating output. That asymmetry is the whole point of the model. New architecture: Causal Encoder-Decoder (CED). 40 layers, 20 causal encoder plus 20 decoder. The decoder’s global KV cache is projected from the encoder’s final hidden states instead of from each decoder layer’s own. Result: a global KV cache of 890 bytes per token. One quarter of V4-Flash, and 1/437 of DeepSeek-V1. Natively multimodal: images go through DeepSeek-ViT, a vision encoder trained from scratch, from the very start of language-model pre-training. Pre-training corpus is 45T tokens. Beats Claude Opus 5 on several agentic benchmarks: 90.6 vs 89.1 on Terminal-Bench 2.1, 74.2 vs 74.0 on DeepSWE v1.1, 54.8 vs 50.3 on AutomationBench. Reasoning effort is a continuous 1-100 dial, not an on/off thinking toggle. API pricing is $0.15 in / $0.60 out per million tokens off-peak, double that at peak. Roughly one twentieth of Claude Opus 5’s output price. DeepSeek is retiring V4 Pro: from September 14, 2026, deepseek-v4-pro requests get routed to V4.1 Flash and billed at Flash rates. Everyone in the open-weight race this year is chasing the same two numbers: active parameters and KV cache. The first decides what each token costs, the second decides how much memory a long context eats. GLM-5.3-Flash answered with hybrid attention, Qwen3.8-Flash-Next with 6B active parameters. ...