← the radar · ✳ model tracker

DeepSeek V4

● live

DeepSeekopen weightsChinese labs stream

Live; open weights; superseded by V4.1 Flash — per DeepSeek's own changelog rather than our own reading, deepseek-v4-flash routes to V4.1 Flash; the matching deepseek-v4-pro cutover set for 04:00 UTC Sep 14 was reversed on its eve and V4 Pro is still served at unchanged billing, which is search-grade too — but our own bench refresh still scored deepseek-v4-pro-0813 as a separate model on Sep 16; our sweep read the bare deepseek-flash alias on Sep 10 at $0.15/$0.60 per 1M, 1M ctx. On the board as live since 2026-04-24 — the watchers keep tracking it: lineup changes, pricing moves and deprecations all alert.

Answering in 0.7s· checked 3h agoIs DeepSeek down? →

Watch for DeepSeek’s next model, free →Launch report (PDF)

Timeline

2026-08-13

DeepSeek V4-Pro lands at $0.435/$0.87 — and "model names remain unchanged" hides a 2.7x price forkpost

DeepSeek launched V4-Pro today on app, web and API. The headline features are agent upgrades, selectable reasoning effort across V4-Pro and V4-Flash, and native OpenAI Responses API support optimised for Codex. The line worth reading twice is the quiet one: "Model names remain unchanged."

2026-07-31

DeepSeek ships V4-Flash official — Terminal-Bench 2.1 jumps to 82.7, and it's $0.14/$0.28confirmed

2026-04-24

DeepSeek V4 verified live — from the vendor's own announcement and catalog listingverified

On the bench index

LMArena: 1,581.9LiveBench: 77.4SimpleBench: 50.9Vals AI: 53Artificial Analysis: 23.1AW Index: 30.4

full comparison table →

Coding leaderboard

#27 of 40 on the coding index, at 34.3 · combines LMArena · Coding, LiveBench · Coding, Vals Vibe Code Bench

the AI coding leaderboard →

Open weights

#5 of 13 open-weights models on the Bench Index · deepseek-ai/DeepSeek-V4-Pro-0813 on Hugging Face · 1.65T parameters · license MIT

the open-weights leaderboard →

The official word

We alert on verified API behavior, never on announcements alone — but when DeepSeek speaks, it's here: deepseek news ↗

Where to use DeepSeek V4

79 providers list DeepSeek V4 in the models.dev catalog (as of 2026-10-27), under the model ids shown below.

DeepSeekdeepseek-v4-pro

Azuredeepseek-v4-pro

Azure Cognitive Servicesdeepseek-v4-pro

OpenRouterdeepseek/deepseek-v4-pro · deepseek/deepseek-v4-pro-0813

Vercel AI Gatewaydeepseek/deepseek-v4-pro · deepseek/deepseek-v4-pro-0813

Abacusdeepseek-ai/DeepSeek-V4-Pro

above.devdeepseek-v4-pro

AI/ML APIdeepseek/deepseek-v4-pro · deepseek/deepseek-v4-pro-0813

ai&deepseek-ai/deepseek-v4-pro

AIHubMixalicloud-deepseek-v4-pro · deep-deepseek-v4-pro · deepseek-v4-pro-0813

Alibaba (China)deepseek-v4-pro

Alibaba Token Plandeepseek-v4-pro · deepseek-v4-pro-0813

Also listed by Alibaba Token Plan (China), AnyAPI, Arcee, Auriko, Baseten, Bothub, Boundless, Charm Hyper, Chutes, ClinePass, Cloudflare AI Gateway, Cloudflare Workers AI, CommandCode, CoreWeave, Cortecs, CrossModel, Deep Infra, DevPass (LLM Gateway), DigitalOcean, EBCloud, Eden AI, EmpirioLabs AI, FastRouter, FrogBot, GMI Cloud, HPC-AI, Hugging Face, Impossibl, IteraCompute, Jalapeno Cloud, Kenari, Kilo Gateway, LLM Gateway, Merge Gateway, Moark, Model Oracle AI, Modelis, NanoGPT, Nebius Token Factory, Novita AI, Ofox, Ollama Cloud, OpenCode, OpenCode Go, Opper, OrcaRouter, Perplexity Agent, Poe, Prime Inference, Requesty, routing.run, RunInfra, SCNet Token Plan, SenseNova (China), SiliconFlow, SiliconFlow (China), Standard Compute, Tempr Gateway, TensorX, Together AI, TokenGo, UnoRouter, Venice AI, Vivgrid, Volcengine Ark, Volcengine Ark Coding Plan, ZenMux.

Listings cover the model family, so a provider may list a sibling tier under the same name. Check the id before you build on it.

DeepSeek V4: quick answers

Is DeepSeek V4 out?

Yes. DeepSeek V4 is live and has been since April 24, 2026.

When was DeepSeek V4 released?

April 24, 2026.

Is DeepSeek V4 good for coding?

On the ArtificialWatch coding index it scores 34.3, #27 of 40 ranked models. The index combines LMArena · Coding, LiveBench · Coding and Vals Vibe Code Bench, taking the median of each model's percentile across them.

Is DeepSeek V4 open source?

It has open weights: the trained model is public on Hugging Face as deepseek-ai/DeepSeek-V4-Pro-0813, 1.65T parameters as Hugging Face counts them, license MIT. Among the 13 open-weights models the ArtificialWatch Bench Index ranks, it is #5.

Who makes DeepSeek V4?

DeepSeek. ArtificialWatch also tracks DeepSeek V4.1 Flash, DeepSeek Code, DeepSeek R3, DeepSeek V4.1 Pro, DeepSeek V4.2, DeepSeek V5 and Next DeepSeek Flash from DeepSeek.

Where can I use DeepSeek V4?

79 providers list it in the models.dev catalog, including DeepSeek, Azure, Azure Cognitive Services, OpenRouter and Vercel AI Gateway. The exact model ids each one lists are on this page.

How much does DeepSeek V4 cost?

DeepSeek's own API lists DeepSeek V4 Pro (deepseek-v4-pro) at $0.66 per million input tokens and $1.98 per million output tokens, and $0.022 per million cached input tokens. Prices as published in the models.dev catalog, as of 2026-10-27.