Skip to content

DeepSeek · Chat

DeepSeek V4 Pro

DeepSeek 🇨🇳ChatDeepSeek familyOpen weights. One of its sources says the weights are publicly available.The maker currently lists 20 Feb 2028 as the deprecation date. Plans can change.
From Empiriolabs's listing:
Flagship MoE LLM with 1.6T total / 49B active parameters and native 1M context for advanced math, logical inference, and specialized coding.
llminfo IQ
130.7
Max context
1.1M
Listings
157

Evidence and sources

🤗 deepseek-ai/DeepSeek-V4-Pro Model architecture, as its sources state it llminfo IQ 130.7, ranked among chat models. From 303 benchmarks. Plenty of data behind it.How llminfo IQ is computedCoding IQ 104.6, 95% interval 98–111. From 22 benchmark families across 14 sources. Plenty of data behind it. Uses the model's best published results. It has its own scale, separate from llminfo IQ.How Coding IQ is computedPoor value (46.9). Costs more per task than its benchmark results predict. From 53 benchmarks with measured task costs. Plenty of data behind it.How Value for money worksYap score: tokens used per coding task vs. models with the same resultsTerse (43 ±10). Generates fewer tokens than models with similar coding results. Measured at default effort. A fair amount of data behind it. By effort: default 43 · none 53 · max 59.Median 21,822 output tokens per task at default effort.How the Yap score worksReleased 24 Apr 2026Maker HQ: CN (according to DeepSeek's own website)
All rankings: #40 on OpenRouter today #40 on OpenRouter this month #45 on OpenRouter this week #95 on NanoGPT recentlySources: DeepSeek data, EmpirioLabs data, LiteLLM data, Models.dev data, Morpheus Inference Marketplace data, llminfo researchPrimary data from DeepSeek from EmpirioLabs: primary mode, description from LiteLLM: deprecation date from Models.dev: weights availability from Morpheus Inference Marketplace: architecture from llminfo research: release date
Tags from its sources:instructtoolsreasoninglong-contextfunction-calling

Advantages and trade-offs

Compared with 10 chat models, picked mainly by popularity and IQ

Advantages

We only make a claim when there's enough data on models of the same type. Updated 22 Nov, 06:07 UTC.

Input & output

as stated by the model's providers
Text
Text

Capabilities

Supported operations and provider-stated features

VisionReasoningFunction callingStructured outputPrompt cachingWeb search

Listings

 DeepSeekdeepseek-v4-proSources: DeepSeek data, GenAI Prices data, LiteLLM data, Models.dev data1M$0.435cache read $0.0036$0.87
 Unorouterdeepseek-v4-pro:free · freeSource: Models.dev data1M$0.00$0.00
 WaferDeepSeek-V4-ProSources: Models.dev data, Wafer data1M$0.28cache read $0.04$3.96
 Routing Rundeepseek-v4-proSource: Models.dev data1M$0.348$0.696
 Apikodeepseek-v4-proSource: Models.dev data1M$0.405cache read $0.0135$1.2149
 EbcloudDeepSeek-V4-ProSource: Models.dev data1M$0.4286$0.8571
 Hugging Facedeepseek-ai/DeepSeek-V4-ProSource: Models.dev data1M$0.435cache read $0.0036$0.87
 Modelisdeepseek-v4-proSource: Models.dev data1M$0.435$0.87
 Tokengodeepseek/deepseek-v4-proSource: Models.dev data1M$0.435$0.87
 Writerdeepseek/deepseek-v4-proSource: Writer PricingNot published by the provider$0.44$0.87

15 listings without a price are hidden.

Included in plans

Benchmarks

Named effort levels run from lowest to highest. Select a result to see its source details.

Human preference

Arena ELO52 related benchmarks · 92 results
Source: LMArena data · Nov 2026
Arena ELO
Chinese
Creative Writing
English
Exclude Ties
Expert
Show 80 more resultsHide 80 results
French
German
Hard Prompts
Hard Prompts English
Industry Entertainment and Sports and Media
Industry Life and Physical and Social Science
Industry Mathematical
Industry Medicine and Healthcare
Industry Software and It Services
Industry Writing and Literature and Language
Japanese
Korean
Longer Query
Multi Turn
Non English
Polish
Russian
Spanish
Style-Controlled
Text Factuality
Text Factuality - Creative Writing
Text Factuality - English
Text Factuality - Exclude Ties
Text Factuality - Expert
Text Factuality - French
Text Factuality - German
Text Factuality - Hard Prompts
Text Factuality - Hard Prompts English
Text Factuality - Industry Software and It Services
Text Factuality - Korean
Text Factuality - Longer Query
Text Factuality - Multi Turn
Text Factuality - Non English
Text Factuality - Russian
Text Factuality - Spanish
Design Arena EloSource: Design Arena Leaderboards data · Nov 2026

Coding

SWE-bench Verified14 related benchmarks · 26 results
SWE-bench Verified
Source: Hugging Face Official Benchmarks data · Reported by the model's developer. The source relaying it did not verify it. claim
SWE-bench Multilingual
Source: Hugging Face Official Benchmarks data · Reported by the model's developer. The source relaying it did not verify it. claim
SWE-bench-Live MultiLang C/C++
Source: SWE-bench-Live leaderboard data · Jun 2026
By scaffold · set: ccpp · instances: 66
SWE-bench-Live MultiLang C#
Source: SWE-bench-Live leaderboard data · Jun 2026
By scaffold · set: csharp · instances: 78
SWE-bench-Live MultiLang Go
Source: SWE-bench-Live leaderboard data · Jun 2026
By scaffold · set: go · instances: 98
Show 14 more resultsHide 14 results
SWE-bench-Live MultiLang Java
Source: SWE-bench-Live leaderboard data · Jun 2026
By scaffold · set: java · instances: 102
SWE-bench-Live MultiLang Rust
Source: SWE-bench-Live leaderboard data · Jun 2026
By scaffold · set: rust · instances: 64
SWE-bench-Live MultiLang TypeScript/JavaScript
Source: SWE-bench-Live leaderboard data · Jun 2026
By scaffold · set: tsjs · instances: 204
SWE-bench-Live Windows
Source: SWE-bench-Live leaderboard data · Jun 2026
scaffold: win-agent, set: windows, instances: 60
SWE-rebench 2026-03
Source: SWE-rebench
SWE-rebench 2026-04
Source: SWE-rebench
SWE-rebench 2026-05
Source: SWE-rebench
SWE-rebench 2026-06
Source: SWE-rebench
ValsSource: Vals AI

Agentic

GAIA4 results
Source: GAIA Leaderboard data · May 2026
Difficulty level
Difficulty not reported
MCPMark VerifiedSource: MCPMark Verified leaderboard docs
OpenHandsSource: OpenHands Index data
Source: OpenRouter benchmark exports
23 Apr 2026 version · 90-day results
Across 20 providers lowest 72% median 75.8% highest 77.1%OpenRouter's automatic routing 77%
Show results from 20 providers and OpenRouter's automatic routingHide results from 20 providers and OpenRouter's automatic routing
  • DeepInfra
  • NextBit
  • SiliconFlow
  • OpenRouter's automatic routing
  • Reka
  • Cloudflare
  • GMICloud
  • Baidu
  • Novita
  • Azure
  • Relace
  • Alibaba
  • Parasail
  • Venice
  • BaseTen
  • StreamLake
  • AtlasCloud
  • DigitalOcean
  • Crusoe
  • Nebius
  • Ionstream

General

Science

GPQA Diamond24 results
Source: Epoch AI Benchmarking Hub data
Reasoning effort
Source: OpenRouter benchmark exports
23 Apr 2026 version · 90-day results
Across 20 providers lowest 54.3% median 88.7% highest 91.7%OpenRouter's automatic routing 88%
Show results from 20 providers and OpenRouter's automatic routingHide results from 20 providers and OpenRouter's automatic routing
  • Cloudflare
  • Novita
  • GMICloud
  • Alibaba
  • Nebius
  • Crusoe
  • DigitalOcean
  • NextBit
  • Ionstream
  • Relace
  • StreamLake
  • Parasail
  • AtlasCloud
  • Baidu
  • OpenRouter's automatic routing
  • Azure
  • Reka
  • SiliconFlow
  • BaseTen
  • Venice
  • DeepInfra

Multilingual

EKI EstonianSource: Keelemudelite mõõdupuu data docs

Long context

Reasoning

Math

Factuality

Instruction following

Vision

Safety

PhareSource: Phare

Knowledge

Other

Nejumi 428 related benchmarks · 56 results
ALT_バイアス
Source: Nejumi Leaderboard 4 data docs
deepseek/deepseek-v4-pro: thinking-max
deepseek-ai/DeepSeek-V4-Pro: reasoning-max
ALT_倫理・道徳
Source: Nejumi Leaderboard 4 data docs
deepseek/deepseek-v4-pro: thinking-max
deepseek-ai/DeepSeek-V4-Pro: reasoning-max
ALT_制御性
Source: Nejumi Leaderboard 4 data docs
deepseek/deepseek-v4-pro: thinking-max
deepseek-ai/DeepSeek-V4-Pro: reasoning-max
ALT_堅牢性
Source: Nejumi Leaderboard 4 data docs
deepseek/deepseek-v4-pro: thinking-max
deepseek-ai/DeepSeek-V4-Pro: reasoning-max
ALT_毒性
Source: Nejumi Leaderboard 4 data docs
deepseek/deepseek-v4-pro: thinking-max
deepseek-ai/DeepSeek-V4-Pro: reasoning-max
ALT_真実性
Source: Nejumi Leaderboard 4 data docs
deepseek/deepseek-v4-pro: thinking-max
deepseek-ai/DeepSeek-V4-Pro: reasoning-max
Show 44 more resultsHide 44 results
AVG_jaster_0shot
Source: Nejumi Leaderboard 4 data docs
deepseek/deepseek-v4-pro: thinking-max
deepseek-ai/DeepSeek-V4-Pro: reasoning-max
AVG_jaster_2shots
Source: Nejumi Leaderboard 4 data docs
deepseek/deepseek-v4-pro: thinking-max
deepseek-ai/DeepSeek-V4-Pro: reasoning-max
GLP_アプリケーション開発
Source: Nejumi Leaderboard 4 data docs
deepseek/deepseek-v4-pro: thinking-max
deepseek-ai/DeepSeek-V4-Pro: reasoning-max
GLP_コーディング
Source: Nejumi Leaderboard 4 data docs
deepseek/deepseek-v4-pro: thinking-max
deepseek-ai/DeepSeek-V4-Pro: reasoning-max
GLP_一般的知識
Source: Nejumi Leaderboard 4 data docs
deepseek/deepseek-v4-pro: thinking-max
deepseek-ai/DeepSeek-V4-Pro: reasoning-max
GLP_基礎的言語性能
Source: Nejumi Leaderboard 4 data docs
deepseek/deepseek-v4-pro: thinking-max
deepseek-ai/DeepSeek-V4-Pro: reasoning-max
GLP_専門的知識
Source: Nejumi Leaderboard 4 data docs
deepseek/deepseek-v4-pro: thinking-max
deepseek-ai/DeepSeek-V4-Pro: reasoning-max
GLP_応用的言語性能
Source: Nejumi Leaderboard 4 data docs
deepseek/deepseek-v4-pro: thinking-max
deepseek-ai/DeepSeek-V4-Pro: reasoning-max
GLP_情報検索
Source: Nejumi Leaderboard 4 data docs
deepseek/deepseek-v4-pro: thinking-max
deepseek-ai/DeepSeek-V4-Pro: reasoning-max
GLP_意味解析
Source: Nejumi Leaderboard 4 data docs
deepseek/deepseek-v4-pro: thinking-max
deepseek-ai/DeepSeek-V4-Pro: reasoning-max
GLP_抽象的推論
Source: Nejumi Leaderboard 4 data docs
deepseek/deepseek-v4-pro: thinking-max
deepseek-ai/DeepSeek-V4-Pro: reasoning-max
GLP_推論能力
Source: Nejumi Leaderboard 4 data docs
deepseek/deepseek-v4-pro: thinking-max
deepseek-ai/DeepSeek-V4-Pro: reasoning-max
GLP_数学的推論
Source: Nejumi Leaderboard 4 data docs
deepseek/deepseek-v4-pro: thinking-max
deepseek-ai/DeepSeek-V4-Pro: reasoning-max
GLP_構文解析
Source: Nejumi Leaderboard 4 data docs
deepseek/deepseek-v4-pro: thinking-max
deepseek-ai/DeepSeek-V4-Pro: reasoning-max
GLP_知識・質問応答
Source: Nejumi Leaderboard 4 data docs
deepseek/deepseek-v4-pro: thinking-max
deepseek-ai/DeepSeek-V4-Pro: reasoning-max
GLP_翻訳
Source: Nejumi Leaderboard 4 data docs
deepseek/deepseek-v4-pro: thinking-max
deepseek-ai/DeepSeek-V4-Pro: reasoning-max
GLP_表現
Source: Nejumi Leaderboard 4 data docs
deepseek/deepseek-v4-pro: thinking-max
deepseek-ai/DeepSeek-V4-Pro: reasoning-max
GLP_論理的推論
Source: Nejumi Leaderboard 4 data docs
deepseek/deepseek-v4-pro: thinking-max
deepseek-ai/DeepSeek-V4-Pro: reasoning-max
GLP_関数呼び出し
Source: Nejumi Leaderboard 4 data docs
deepseek/deepseek-v4-pro: thinking-max
deepseek-ai/DeepSeek-V4-Pro: reasoning-max
TOTAL_SCORE
Source: Nejumi Leaderboard 4 data docs
deepseek/deepseek-v4-pro: thinking-max
deepseek-ai/DeepSeek-V4-Pro: reasoning-max
アラインメント(ALT)_AVG
Source: Nejumi Leaderboard 4 data docs
deepseek/deepseek-v4-pro: thinking-max
deepseek-ai/DeepSeek-V4-Pro: reasoning-max
汎用的言語性能(GLP)_AVG
Source: Nejumi Leaderboard 4 data docs
deepseek/deepseek-v4-pro: thinking-max
deepseek-ai/DeepSeek-V4-Pro: reasoning-max
Vals10 related benchmarks · 14 results
EMB
Source: Vals AI
CaseLaw v2
Source: Vals AI
CorpFin v2
Source: Vals AI
LegalBench
Source: Vals AI
MedCode
Source: Clinical Benchmarks data docs · Sep 2026
Setting not stated
Setting not stated
Source: Vals AI
Setting not stated
Show 6 more resultsHide 6 results
MedScribe
Source: Clinical Benchmarks data docs · Sep 2026
Setting not stated
Setting not stated
Source: Vals AI
Setting not stated

Data credits: Hugging Face official benchmark leaderboards; licence varies by benchmark · SEA-HELM leaderboard, AI Singapore (AI Products team), National University of Singapore. Published aggregate scores; results licence not stated (SEA-HELM code is MIT). · Epoch AI Benchmarking Hub, CC BY 4.0 (licence) · Clinical Benchmarks (clinicalbenchmarks.ai), CC BY 4.0; fractions converted to percentages; source documents retain their own terms (licence) · LMArena, CC BY 4.0 (licence) · Baromètre de l’écriture IA, Johannes Eva; public numeric ranking data, © 2026 Johannes Eva, no results licence stated · BullshitBench by Peter Gostev, MIT (licence) · EQ-Bench (Samuel J. Paech) — cite paech2023eqbench · CyberGym leaderboard (Wang, Shi, He, Cai, Zhang, Song; UC Berkeley), ICLR 2026. Results evaluated and submitted by the listed teams. No licence is stated for the result data; the repository's Apache-2.0 licence covers its code only. · Keelemudelite mõõdupuu: Independent LLM leaderboard for Estonian. Eesti Keele Instituut. https://moodupuu.eki.ee, CC BY 4.0. Latest per-benchmark results selected with partial-run flags retained. (licence) · OpenRouter public benchmark leaderboard and search-lane exports, CC BY 4.0. Fractional source scores converted to percentages; immutable dated snapshots. (licence) · HalluHard: Fan, Delsad, Flammarion and Andriushchenko (EPFL), arXiv:2602.01031 · MCPMark (eval-sys), mcpmark.ai Verified leaderboard. Single-run Pass@1 fractions shown as percentages. · MineBench, MIT (licence) · Nejumi Leaderboard 4, Weights & Biases Japan; public publisher-selected numeric evaluation results · OpenHands Index (All Hands AI), Apache-2.0 (licence) · Phare (Giskard), CC BY 4.0 (licence) · SnakeBench (Greg Kamradt / ARC Prize Foundation), MIT (licence) · SWE-bench-Live (Microsoft and contributors), swe-bench-live.github.io. Resolve rates computed from the published report counts. · UzLiB, Tahrirchi (Abror Shopulatov). Public LEADERBOARD.md numbers; repository MIT covers software and documentation, no separate results licence stated and none asserted. · Vectara Hallucination Leaderboard, Apache-2.0 (licence)

Also known as: aihubmix/deepseek-v4-pro, alibaba/deepseek-v4-pro, azure_ai/deepseek-v4-pro, azure_ai/fw-deepseek-v4-pro, baseten/deepseek-ai/deepseek-v4-pro, cline-pass/deepseek-v4-pro, dashscope/deepseek-v4-pro, deepinfra/deepseek-ai/deepseek-v4-pro, deepinfra/deepseek-v4-pro, deepinfra/deepseek-v4-pro-0424, deepseek-ai/deepseek-v4-pro, deepseek-v4-pro-260425, deepseek-v4-pro-fast, deepseek-v4-pro-high-0813, deepseek-v4-pro-high-20260813, deepseek-v4-pro:free, deepseek-v4-pro:web, deepseek-v4-pro@eu, deepseek/deepseek-v4-pro, deepseek/deepseek-v4-pro-0423, deepseek/deepseek-v4-pro-20260423, deepseek/deepseek-v4-pro-20260423:thinking, deepseek/deepseek-v4-pro-202606, empiriolabs/deepseek-v4-pro, fireworks_ai/accounts/fireworks/models/deepseek-v4-pro, nebius/deepseek-ai/deepseek-v4-pro, novita/deepseek/deepseek-v4-pro, opencode_go/deepseek-v4-pro, openrouter/deepseek/deepseek-v4-pro, qwen_ai_platform/deepseek-v4-pro, qwencloud/deepseek-v4-pro, runware/deepseek-v4-pro, streamlake/deepseek-v4-pro, tencent/deepseek-v4-pro, tensorx/deepseek-v4-pro-0424, together-ai/deepseek-v4-pro, wandb/deepseek-ai/deepseek-v4-pro

Like this one. Tap or hover any dotted value, label or icon to see where it comes from and how it is measured. open details when tapped or hovered. How this data works