跳轉到主要內容

DeepSeek V4 Pro 0813 (Max)

DeepSeekDeepSeek開源權重MIT · 商用許可

描述

DeepSeek-V4-Pro-0813 is the GA / production release of DeepSeek-V4-Pro, with major upgrades for agentic coding and tool-use workloads. It is served behind the official API id `deepseek-v4-pro` and is distinct from catalog `deepseek-v4-pro-max`, which is the maximum reasoning-effort mode of Pro rather than this dated GA build. Same Pro MoE scale as peers (1.6T total / 49B activated). It supports low, high, and max reasoning effort, a 1M-token context window, up to 384K output tokens, and DeepSeek's OpenAI-compatible Chat Completions and Responses APIs. GA open weights are on Hugging Face at `deepseek-ai/DeepSeek-V4-Pro-0813` (MIT), distinct from the April preview repo `deepseek-ai/DeepSeek-V4-Pro`. New time-of-day pricing takes effect August 16: peak rates are $0.044 cached input, $1.32 uncached input, and $3.96 output per million tokens; off-peak rates are half those amounts.

發布日期
2026-08-13
參數規模
1.6T
上下文長度
1.0M
支援模態
text

能力雷達圖

23
general
70
coding
50
reasoning
43
science估算
50
agents
0
multimodal

缺少專門科學評測時,Science 由 LLM Stats 科學得分或推理能力估算。

排行榜排名

領域#排名分數來源
智慧體能力模型榜100
42.0
LS
通用能力榜42
34.0
AA
數學推理25
91.0
LB
科學能力41
36.0
AA

基準測試分數 (LLM Stats)

(LLM Stats (zeroeval))

Agents

Toolathlon51.8%自報
AutomationBench31.8%自報
Agents' Last Exam25.7%自報

Code

CyberGym83.3%自報
DSBench-FullStack71.1%自報
DSBench-Hard67.2%自報
DeepSWE62.7%自報
NL2Repo61.5%自報
FrontierSWE29.0%

Factuality

SimpleQA57.9%自報

General

CSimpleQA84.4%自報

Language

MMLU-Pro87.5%自報

Long Context

MRCR 1M83.5%自報
CorpusQA 1M62.0%自報

Math

CodeForces1.00 / 3000自報
HMMT Feb 2695.2%自報
MathArena Apex90.2%自報
IMO-AnswerBench89.8%自報
LiveBench73.6%

Reasoning

LiveCodeBench93.5%自報
GPQANYU + Cohere + Anthropic (2023)90.1%自報
Terminal-Bench 2.187.9%自報
BrowseCompOpenAI (2025)83.4%自報
SWE-Bench Verified80.6%自報
SWE-bench Multilingual76.2%自報
MCP Atlas73.6%自報
Terminal-Bench 2.0Stanford × Laude Institute (2026)67.9%自報
Humanity's Last Exam (with tools, text-only)60.0%自報
SWE-Bench ProPrinceton NLP (2024)55.4%自報
Humanity's Last Exam48.2%自報
Humanity's Last Exam (no tools, text-only)42.7%自報
FrontierCode 1.117.6%

AA 評測指數

(Artificial Analysis)
Strategy And Ops
41.1
Finance And Accounting
30.4
Engineering(Artificial Analysis)
29.8
Economics
29.3
Legal
27.5
Intelligence Index(Artificial Analysis)
23.1

LLM Stats 分類評分

(LLM Stats (zeroeval))
Safety
80
Code
70
General
60
Agents
60
Vision
60
Math
50
Reasoning
50
Tool Calling
50

定價

輸入價格$1.32 / 1M tokens
輸出價格$3.96 / 1M tokens
混合價格(3:1)$1.625 / 1M tokens
快取讀取價格$0.022 / 1M tokens

速度

Tokens/秒93.5
首Token延遲1.49s
首回答延遲22.88s

供應商價格排行

供應商價格排行

60 個供應商

最便宜: routing.run最貴: Adzuki AI
供應商輸入輸出
1routing.run最便宜
$0.348
$0.696
2APIKO
$0.405
$1.2149
3Modelis
$0.435
$0.87
4TokenGo
$0.435
$0.87
5JZSTOKEN
$0.4491
$2.25
6Kenari
$0.5714
$1.14
7OpenCode Go
$0.66
$1.98
8ZenMux
$0.66
$1.98
9Ollama Cloud
$0.66
$1.98
10DevPass (LLM Gateway)
$0.66
$1.98
11OrcaRouter
$0.66
$1.98
12Eden AI
$0.66
$1.98
13LLM Gateway
$0.66
$1.98
14CommandCode
$0.66
$1.98
15TeamoRouter
$0.66
$1.98
16Ezkielyna
$0.66
$1.976
17Tokenify
$0.66
$1.98
18above.dev
$0.726
$2.178
19UnoRouter
$0.8999
$1.7999
20OpenRouter
$0.9
$1.5
21AI/ML API
$0.9
$2.7
22Standard Compute
$1.044
$2.088
23DeepInfra
$1.3
$2.6
24DeepSeek主要
$1.32
$3.96
25Together
$1.32
$3.96
26Novita
$1.32
$3.96
27FrogBot
$1.32
$3.96
28Requesty
$1.32
$3.96
29Vercel AI Gateway
$1.32
$3.96
30Auriko
$1.32
$3.96
31Merge Gateway
$1.32
$3.96
32Ofox
$1.32
$3.96
33SandBase
$1.32
$3.96
34Synthorai
$1.32
$3.96
35CrossModel
$1.35
$4.05
36RuAPI
$1.584
$4.752
37Novita AI
$1.6
$3.2
38Kilo Gateway
$1.6
$3.2
39Alibaba (China)
$1.65
$3.301
40Venice AI
$1.65
$3.301
41EmpirioLabs AI
$1.65
$3.3
42DotHoosh
$1.694
$5.082
43Cortecs
$1.7
$3.4
44CloudCode.ONE
$1.705
$5.115
45Vivgrid
$1.74
$3.48
46Cloudflare AI Gateway
$1.74
$3.48
47OpenCode
$1.74
$3.48
48Azure Cognitive Services
$1.74
$3.48
49DigitalOcean
$1.74
$3.48
50Azure
$1.74
$3.48
51FastRouter
$1.74
$3.48
52HPC-AI
$1.74
$3.48
53Impossibl
$1.74
$3.48
54Arcee
$1.74
$3.48
55TensorX
$1.75
$3.5
56Opper
$1.78812
$3.57624
57Gisoo
$1.848
$5.544
58NaN
$1.85328
$5.55984
59Prime Inference
$1.91
$3.83
60Adzuki AI
$2.5714
$7.7143

比較該模型在不同 API 供應商之間的定價。

外部連結