40 个模型

DeepSeek 模型：智能、性能与价格

Name: Artificial Analysis Intelligence Index
Creator: Artificial Analysis
License: https://artificialanalysis.ai/docs/legal/Terms-of-Use.pdf

DeepSeek

智能

Updated

DeepSeek V4.1 Flash 1017 (max)

DeepSeek V4.1 Flash 1017 (high)

DeepSeek V4 Pro 0813 (max)

DeepSeek V4.1 Flash 1017 (non-reasoning)

DeepSeek V4 Pro 0813 (non-reasoning)

Artificial Analysis Intelligence Index

每任务耗时

DeepSeek V4.1 Flash 1017 (non-reasoning)

7.5 分钟

DeepSeek V4.1 Flash 1017 (high)

11.2 分钟

DeepSeek V4.1 Flash 1017 (max)

14.4 分钟

DeepSeek V4 Pro 0813 (non-reasoning)

15.5 分钟

DeepSeek V4 Pro 0813 (max)

24.0 分钟

每项 Intelligence Index 任务

成本

DeepSeek V4.1 Flash 1017 (non-reasoning)

$0.43

DeepSeek V4.1 Flash 1017 (high)

$0.46

DeepSeek V4.1 Flash 1017 (max)

$0.60

DeepSeek V4 Pro 0813 (non-reasoning)

$0.98

DeepSeek V4 Pro 0813 (max)

$1.77

每项 Intelligence Index 任务

Artificial Analysis 已对 DeepSeek 的 40 个模型进行基准测试。下方对这些模型的关键指标进行了比较。

智能方面，DeepSeek 得分最高的模型是 DeepSeek V4.1 Flash 1017 (Max)（33）。
速度方面，DeepSeek V4.1 Flash 1017 (Non-reasoning)（7.5 分钟）的每任务耗时最短。
延迟方面，DeepSeek V4.1 Flash 1017 (Non-reasoning)（1.31 秒）的首个答案 token 等待时间最短。
价格方面，DeepSeek V4.1 Flash 1017 (Non-reasoning)（$0.43）的单任务成本最低。不同模型的价格差异最高可达 4.2 倍。

智能Updated

Artificial Analysis Intelligence Index

Artificial Analysis Intelligence Index v5.0.1 incorporates 9 evaluations: AA-Briefcase v1.1, AutomationBench-AA, Terminal-Bench 4.0, FrontierCode 2.0, AA-Omniscience v1.1, GDP.pdf, HLE-Diamond, CritPt v2 (Hard Subset), and Terminal-Bench-Science 0.1

Artificial Analysis Intelligence Index v5.0.1 includes: AA-Briefcase v1.1, AutomationBench-AA, Terminal-Bench 4.0, FrontierCode 2.0, AA-Omniscience v1.1, GDP.pdf, HLE-Diamond, CritPt v2 (Hard Subset), Terminal-Bench-Science 0.1. See Intelligence Index methodology for further details, including a breakdown of each evaluation and how we run them.

Intelligence Index 与每项任务的成本

Artificial Analysis Intelligence Index · Weighted average cost (USD) per Artificial Analysis Intelligence Index task

Most attractive quadrant

Pareto line

Weighted average cost per Intelligence Index task. Each evaluation’s cost is calculated from input, cache hit, cache write, reasoning, and answer token prices, divided by task count, and weighted by its Intelligence Index weight.

成本

每项 Intelligence Index 任务的成本

Weighted average cost (USD) per Intelligence Index task · Segmented by token type · Lower is better

速度与延迟

Intelligence Index 每项任务耗时

Output tokens divided by end-to-end output speed (minutes per Intelligence Index task) · Speed at the model's own endpoint or the median across providers · Lower is better

The weighted average time (minutes) per Artificial Analysis Intelligence Index task: output tokens per task divided by output speed, weighted by each benchmark's weight in the Intelligence Index.

Speed and latency metrics average each endpoint's runs over the past 5 days, and the error rate pools them. We run the benchmark every 6 hours, and endpoints of deprecated models every 18 hours.

能力得分

能力指数

衡量模型在特定能力与行业中的表现

Finance & Accounting Index

Incorporates 5 evaluations: AA-Briefcase v1.1, AA-Omniscience v1.1, HLE-Diamond, AutomationBench-AA, GDP.pdf · Higher is better

Strategy & Ops Index

Incorporates 4 evaluations: AutomationBench-AA, AA-Briefcase v1.1, AA-Omniscience v1.1, GDP.pdf · Higher is better

Legal Index

Incorporates 5 evaluations: AA-Omniscience v1.1, AA-Briefcase v1.1, HLE-Diamond, GDP.pdf, AutomationBench-AA · Higher is better

Healthcare & Medical Index

Incorporates 5 evaluations: AA-Omniscience v1.1, AA-Briefcase v1.1, MLCR-AA, AutomationBench-AA, HLE-Diamond · Higher is better

Engineering Index

Incorporates 7 evaluations: AA-Omniscience v1.1, CritPt v2 (Hard Subset), Terminal-Bench-Science 0.1, HLE-Diamond, Terminal-Bench 4.0, FrontierCode 2.0, AA-Briefcase v1.1 · Higher is better

Economics Index

Incorporates 4 evaluations: AA-Omniscience v1.1, AA-Briefcase v1.1, HLE-Diamond, GDP.pdf · Higher is better

DeepSeek 的所有发布

更多详情

								权重		服务商基准测试
DeepSeek V4.1 Flash 1017 (Max)	DeepSeek	33	552B 推理时启用 16B 个参数	1M	US$0.2	14.4	161	🤗	+19	查看
DeepSeek V4.1 Flash 1017 (High)	DeepSeek	29	552B 推理时启用 16B 个参数	1M	US$0.2	11.2	164	🤗		查看
DeepSeek V4 Pro 0813 (Max)	DeepSeek	23	1.6T 推理时启用 49B 个参数	1M	US$0.7	24.0	69	🤗	+6	查看
DeepSeek V4.1 Flash 1017 (Non-reasoning)	DeepSeek	19	552B 推理时启用 16B 个参数	1M	US$0.2	7.5	145	🤗	+3	查看
DeepSeek V4 Pro 0813 (Non-reasoning)	DeepSeek	13	1.6T 推理时启用 49B 个参数	1M	US$0.7	15.5	77	🤗		查看
DeepSeek V3.2 Exp (Non-reasoning)	DeepSeek	-	685B 推理时启用 37B 个参数	128k	US$0.1	-	-	🤗		查看
DeepSeek V3.1 Terminus (Reasoning)	DeepSeek	-	685B 推理时启用 37B 个参数	128k	US$1.7	-	-	🤗		查看
DeepSeek R1 (Jan '25)	DeepSeek	-	685B 推理时启用 37B 个参数	128k	US$2.2	-	124	🤗		查看
DeepSeek R1 Distill Llama 8B	DeepSeek	-	8B	128k	-	-	-	🤗	-	查看
DeepSeek V3.1 (Reasoning)	DeepSeek	-	685B 推理时启用 37B 个参数	128k	US$0.5	-	117	🤗		查看
DeepSeek-V2-Chat	DeepSeek	-	236B 推理时启用 21B 个参数	128k	-	-	-	🤗	-	查看
DeepSeek V3 (Dec '24)	DeepSeek	-	671B 推理时启用 37B 个参数	128k	US$0.4	-	74	🤗		查看
DeepSeek R1 Distill Qwen 1.5B	DeepSeek	-	1.5B	128k	-	-	-	🤗	-	查看
DeepSeek R1 Distill Llama 70B	DeepSeek	-	70B	128k	US$0.7	-	-	🤗		查看
DeepSeek V3.2 Speciale	DeepSeek	-	685B 推理时启用 37B 个参数	128k	-	-	-	🤗	-	查看
DeepSeek V4 Pro 0424 (High)	DeepSeek	-	1.6T 推理时启用 49B 个参数	1M	US$0.7	-	-	🤗	+2	查看
DeepSeek R1 0528 Qwen3 8B	DeepSeek	-	8.2B	33k	-	-	-	🤗	-	查看
DeepSeek V3.2 (Non-reasoning)	DeepSeek	-	685B 推理时启用 37B 个参数	128k	US$0.3	-	37	🤗	+5	查看
DeepSeek V4 Pro 0424 (Non-reasoning)	DeepSeek	-	1.6T 推理时启用 49B 个参数	1M	US$0.8	-	161	🤗		查看
DeepSeek V4.1 Flash (Non-reasoning)	DeepSeek	-	552B 推理时启用 16B 个参数	1M	US$0.2	-	221	🤗	+5	查看
DeepSeek V3 0324	DeepSeek	-	671B 推理时启用 37B 个参数	128k	US$0.2	-	155	🤗		查看
DeepSeek V4 Flash Vision (Max)	DeepSeek	-	284B 推理时启用 13B 个参数	1M	US$0.2	-	159	不可用		查看
DeepSeek LLM 67B Chat (V1)	DeepSeek	-	7B	4k	-	-	-	🤗	-	查看
DeepSeek V4 Flash 0420 (High)	DeepSeek	-	284B 推理时启用 13B 个参数	1M	US$0.1	-	-	🤗		查看
DeepSeek-V2.5	DeepSeek	-	236B 推理时启用 21B 个参数	128k	-	-	-	🤗	-	查看
DeepSeek-Coder-V2	DeepSeek	-	236B 推理时启用 21B 个参数	128k	-	-	-	🤗	-	查看
DeepSeek R1 0528 (May '25)	DeepSeek	-	685B 推理时启用 37B 个参数	128k	US$1.8	-	24	🤗		查看
DeepSeek V4 Flash 0420 (Non-reasoning)	DeepSeek	-	284B 推理时启用 13B 个参数	1M	US$0.1	-	-	🤗		查看
DeepSeek V3.2 Exp (Reasoning)	DeepSeek	-	685B 推理时启用 37B 个参数	128k	US$0.1	-	-	🤗		查看
DeepSeek R1 Distill Qwen 14B	DeepSeek	-	14B	128k	-	-	-	🤗	-	查看
DeepSeek-V2.5 (Dec '24)	DeepSeek	-	236B 推理时启用 21B 个参数	128k	-	-	-	🤗	-	查看
DeepSeek V4 Pro 0424 (Max)	DeepSeek	-	1.6T 推理时启用 49B 个参数	1M	US$0.6	-	75	🤗	+4	查看
DeepSeek Coder V2 Lite Instruct	DeepSeek	-	16B 推理时启用 2.4B 个参数	128k	-	-	-	🤗	-	查看
DeepSeek V3.2 (Reasoning)	DeepSeek	-	685B 推理时启用 37B 个参数	128k	US$0.1	-	35	🤗	+3	查看
DeepSeek V4 Flash 0420 (Max)	DeepSeek	-	284B 推理时启用 13B 个参数	1M	US$0.1	-	38	🤗		查看
DeepSeek V4.1 Flash (Max)	DeepSeek	-	552B 推理时启用 16B 个参数	1M	US$0.2	-	211	🤗	+13	查看
DeepSeek R1 Distill Qwen 32B	DeepSeek	-	32B	128k	-	-	-	🤗	-	查看
DeepSeek V3.1 Terminus (Non-reasoning)	DeepSeek	-	685B 推理时启用 37B 个参数	128k	US$0.2	-	68	🤗		查看
DeepSeek V3.1 (Non-reasoning)	DeepSeek	-	685B 推理时启用 37B 个参数	128k	US$0.5	-	72	🤗	+3	查看
DeepSeek V4 Flash 0731 (Max)	DeepSeek	-	284B 推理时启用 13B 个参数	1M	US$0.1	-	123	🤗	+8	查看

DeepSeek 模型：智能、性能与价格

智能

每任务耗时

成本

智能Updated

Artificial Analysis Intelligence Index

Artificial Analysis Intelligence Index

Intelligence Index 与每项任务的成本

Cost per Intelligence Index Task

Artificial Analysis Intelligence Index

成本

每项 Intelligence Index 任务的成本

Cost per Intelligence Index Task

速度与延迟

Intelligence Index 每项任务耗时

Time per Intelligence Index Task

Time Period

能力得分

能力指数

DeepSeek 的所有发布

更多详情