DeepSeek V4.1 Flash 1017 vs. Qwen3.8 27B: comparación de lanzamientos

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

Comparación de los lanzamientos DeepSeek V4.1 Flash 1017 y Qwen3.8 27B: 2 y 4 modelos respectivamente, cada uno con distintas características de inteligencia, rendimiento y precio. A continuación se comparan las métricas clave de los 6 modelos.

En inteligencia, DeepSeek V4.1 Flash 1017 obtiene la puntuación más alta: DeepSeek V4.1 Flash 1017 (Max), con 32, frente a Qwen3.8 27B (Xhigh), con 20 de Qwen3.8 27B.
En velocidad de salida, DeepSeek V4.1 Flash 1017 es el más rápido: DeepSeek V4.1 Flash 1017 (Non-reasoning), con 192 tokens/s, frente a Qwen3.8 27B (Medium), con 48 tokens/s de Qwen3.8 27B.
En costo por tarea, DeepSeek V4.1 Flash 1017 es el más económico: DeepSeek V4.1 Flash 1017 (Non-reasoning), con $0.43, frente a Qwen3.8 27B (Low), con $1.91 de Qwen3.8 27B.

Playground de modelos

Lanzamientos comparados


	DeepSeek V4.1 Flash 1017 (max)	Qwen3.8 27B (xhigh)	DeepSeek V4.1 Flash 1017 (non-reasoning)	Qwen3.8 27B (medium)	Qwen3.8 27B (low)	Qwen3.8 27B (non-reasoning)
Intelligence Index	32	20	19	17	16	10
Finanzas y contabilidad	42	36	37	29	28	22
Estrategia y operaciones	49	39	40	34	29	16
Jurídico	38	33	32	26	27	22
Salud y medicina	41	35	--	24	23	19
Ingeniería	37	28	25	22	21	17
Economía	40	35	34	29	28	22
AA-Briefcase v1.1	1570	1393	1426	1373	1270	901
AutomationBench-AA	66.6%	48.2%	48.8%	40.3%	30.1%	12.4%
Terminal-Bench 4.0	33.8%	5.6%	9.1%	5.1%	2.5%	0.0%
FrontierCode 2.0	28.5%	14.9%	6.1%	11.5%	12.4%	3.7%
AA-Omniscience v1.1	1	−6	2	−36	−21	−7
GDP.pdf	17.2%	16.4%	4.4%	12.8%	10.4%	5.2%
HLE-Diamond	25.3%	15.3%	7.1%	6.7%	6.8%	4.2%
CritPt v2 (Hard Subset)	24.0%	4.0%	1.0%	16.5%	5.5%	2.0%
Terminal-Bench-Science 0.1	12.9%	0.5%	2.4%	1.0%	0.5%	0.0%

Costo por tareaUSD	$0.60	$3.53	$0.43	$2.16	$1.91	$3.77
Precio de entradaUSD/1M tokens	$0.30	$0.50	$0.30	$0.50	$0.50	$0.50
Precio de salidaUSD/1M tokens	$1.20	$3.00	$1.20	$3.00	$3.00	$3.00
Precio de lectura de cachéUSD/1M tokens	$0.006	$0.10	$0.006	$0.10	$0.10	$0.10
Precio de escritura de cachéUSD/1M tokens	--	--	--	--	--	--

Velocidad de salidaTokens/s	173	40	192	48	47	45
Tiempo hasta el primer tokens	1,13	4,02	1,36	4,09	3,90	4,02
Tiempo hasta el primer token de respuestas	12,68	53,65	1,36	45,38	46,10	4,02
Tiempo de respuesta de extremo a extremos	15,56	66,06	3,97	55,70	56,66	15,08
Tiempo por tareas	590,84	1842,60	340,80	1613,55	1514,16	935,32

Ventana de contexto	1M	256k	1M	256k	256k	256k
Creador	DeepSeek	Alibaba	DeepSeek	Alibaba	Alibaba	Alibaba
Licencia	Abierto	Abierto	Abierto	Abierto	Abierto	Abierto
Modalidad de entrada	Admite: texto e imágenes	Admite: texto, imágenes y video	Admite: texto e imágenes	Admite: texto, imágenes y video	Admite: texto, imágenes y video	Admite: texto, imágenes y video
Modalidad de salida	Admite: texto	Admite: texto	Admite: texto	Admite: texto	Admite: texto	Admite: texto
Proveedores	+16	+7				+4

Updated

Inteligencia

Artificial Analysis Intelligence Index · Higher is better

Velocidad

Output tokens per second · Higher is better

Costo por tarea

Weighted average cost (USD) per Intelligence Index task · Lower is better

InteligenciaUpdated

Artificial Analysis Intelligence Index

Artificial Analysis Intelligence Index v5.0 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 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 vs. costo por tarea

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

Most attractive quadrant

DeepSeek V4.1 Flash 1017

Qwen3.8 27B

Pareto line

AA-Briefcase v1.1

AutomationBench-AA

Costo por tarea (USD, escala logarítmica)

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.

Costo

Costo por tarea del Intelligence Index

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

Velocidad y latencia

Velocidad de salida

Output tokens per second · Higher is better

Tokens per second received while the model is generating tokens (ie. after first chunk has been received from the API for models which support streaming).

Figures represent performance of the model's first-party API or the median across providers where a first-party API is not available.

Puntuaciones de capacidades

Índices de capacidades

Mide el rendimiento de los modelos en capacidades e industrias específicas

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

Benchmarks

AA-PersonalityBenchNew

Modelo Avs.Modelo B

AA-PersonalityBench: índice Expressiveness, DeepSeek V4.1 Flash 1017 (max) vs. Qwen3.8 27B (xhigh)

Probabilidad de que la respuesta de cada modelo sea juzgada como la que muestra más cada rasgo, según su diferencia de Elo · 50 % es un empate

DeepSeek V4.1 Flash 1017 (max)

Qwen3.8 27B (xhigh)

Lexically Varied

97 %

3 %

Literate

84 %

16 %

Natural Sounding

66 %

34 %

Eloquent

62 %

38 %

Metaphorical

49 %

51 %

DeepSeek V4.1 Flash 1017 vs. Qwen3.8 27B: comparación de lanzamientos

Inteligencia

Velocidad

Costo por tarea

InteligenciaUpdated

Artificial Analysis Intelligence Index

Artificial Analysis Intelligence Index

Intelligence Index vs. costo por tarea

Cost per Intelligence Index Task

Artificial Analysis Intelligence Index

Costo

Costo por tarea del Intelligence Index

Cost per Intelligence Index Task

Velocidad y latencia

Velocidad de salida

Output Speed

Model Performance Representation

Puntuaciones de capacidades

Índices de capacidades

Benchmarks

AA-PersonalityBenchNew

AA-PersonalityBench: índice Expressiveness, DeepSeek V4.1 Flash 1017 (max) vs. Qwen3.8 27B (xhigh)