Command Palette
Search for a command to run...

DeepSeek Launches Open-Source V4 Models With 1M-Token Context

aiai-modelingai-products 90 posts · 35 accounts

DeepSeek said its open-sourced DeepSeek-V4 Preview is now live, releasing V4 Pro and V4 Flash with a 1 million-token context window and immediate API availability. The company said V4 Pro has 1.6 trillion total parameters with 49 billion active, while V4 Flash has 284 billion total and 13 billion active; both support thinking and non-thinking modes and are compatible with OpenAI ChatCompletions and Anthropic APIs. DeepSeek said the models use token-wise compression and DeepSeek Sparse Attention to reduce compute and memory costs for long-context tasks.

Artificial Analysis said V4 Pro ranked as the top open-weight model on its GDPval-AA benchmark, scoring 1554, ahead of GLM-5.1 at 1535 and Kimi K2.6 at 1484. The firm listed DeepSeek's first-party API pricing at $1.74/$3.48 per 1 million input/output tokens for V4 Pro and $0.14/$0.28 for V4 Flash. Vals AI separately said DeepSeek V4 was the top open-weight model on its Vibe Code Benchmark, though early benchmark results for open models may not fully reflect real-world performance until broader testing.

From the sources (25 posts)

@scaling01

DEEPSEEK-V4 FLASH AND PRO ITS HAPPENING

@scaling01

DeepSeek-V4 Pricing

@scaling01

DeepSeek-V4 with 1M context length and a maximum of 384k output tokens The pricing is still insane Just $3.48 output

@scaling01

DeepSeek-V4 official pricing: DeepSeek-V4 Flash:$0.14 / $0.28 DeepSeek-V4 Pro: $1.74 / $3.48

@teortaxestex

FINALLLY FINALLY it is here. V4-flash: all the way back to V2 prices, only now with 1M V4-pro: roughly Kimi/GLM/MiMo competitor Chat prefix completion and FIM back – thank you! Missed this forever but what can they do?

@mweinbach

OOOOO Deepseek v4 Pro is already working on the API

@scaling01

DeepSeek-V4 Benchmarks HuggingFace:

@scaling01

DeepSeek-V4 parameters confirmed: Flash is 284B@13B Pro is 1.6T@49B

@mweinbach

Deepseek V4 weights are up v4 Flash is a 284B MoE, v4 Pro is 1.6T MoE

@teortaxestex

DeepSeek V4 has maximum output length of 384 thousand tokens.

@mweinbach

Deepseek v4 benchmarks

@scaling01

DeepSeek-V4 was pre-trained on 32T tokens using Muon and integrates a new hybrid attention mechanism and mHC

@jukan05

What kind of magic did DeepSeek pull off this time? With V4, they seem to be back at SOTA again. Their coding performance also looks pretty serious.

@scaling01

DeepSeek-V4 Technical Report

@stevibe

DeepSeek V4 Flash, DeepSeek V4 Pro, from their docs!

@lmsysorg

DeepSeek V4 by @deepseek_ai just dropped! SGLang is ready on Day 0 with a full stack of optimizations from architectures to low-level kernels. We also deliver a verified RL training pipeline in Miles (by @radixark) for V4 at launch: 1️⃣ Na

@kimmonismus

Deepseek v4 pro Evals. Roughly on par with GPT-5.4 xhigh and opus 4.6 max

@fabknowledge

RT @scaling01: DeepSeek-V4 Benchmarks HuggingFace:

@theahmadosman

DeepSeek V4 IS HERE

@scaling01

"DeepSeek-V4-Pro-Max outperforms Opus-4.6-Max on diverse Chinese white-collar tasks, achieving an impressive non-loss rate of 63%"

@zerohedge

*DEEPSEEK UNVEILS PREVIEW VERSIONS OF LATEST V4 AI MODEL here we go again

@scaling01

DeepSeek-V4 seems to be on a level with Opus 4.5 on real world agentic coding tasks

@theahmadosman

FINALLY DeepSeek V4 IS HERE - 2 Versions, Flash and Pro - Each comes in Instruct and Base - Flash is 284B MoE with 13B Activated Parameter per Token - Pro is 1.6T MoE w/ 49B Activated Parameter per Token - 1M Context Length The newest

@valsai

DeepSeek v4 is now the #1 open-weight model on our Vibe Code Benchmark, and it’s not close. It leaves the #2 (Kimi K2.6) in the dust, and even beats out frontier closed source models like Gemini 3.1 Pro.

@poezhao0605

DeepSeek launched V4 today. Two models, both open-source, both with 1M token context windows. V4-Pro: 1.6T total parameters, 49B active. Benchmarks put it alongside Claude Opus 4.6 and GPT-5.4. V4-Flash: 284B parameters, 13B active. Desig

Preview built on a synthetic news corpus (16 weeks, Apr–Jul 2026). Impact calls are model reads, not price data.

About Archive