DeepInfra: inteligencia, rendimiento y precio de sus modelos

DeepInfra
DeepInfra

Análisis de los modelos de DeepInfra en métricas clave como calidad, precio, velocidad de salida, latencia, ventana de contexto y más. Este análisis está pensado para ayudarte a elegir el mejor modelo ofrecido por DeepInfra para tu caso de uso.

Más inteligente

#1
GLM-5.2 (max) (FP4)
GLM-5.2 (max) (FP4)
51
#2
MiniMax-M3
MiniMax-M3
44
#3
DeepSeek V4 Pro (max) (FP4)
DeepSeek V4 Pro (max) (FP4)
44
#4
Kimi K2.6 (FP4)
Kimi K2.6 (FP4)
44
#5
DeepSeek V4 Pro (high) (FP4)
DeepSeek V4 Pro (high) (FP4)
43

Índice de inteligencia

93 modelos en total

Más rápido

#1
gpt-oss-120b (high) (Turbo)
gpt-oss-120b (high) (Turbo)
253 t/s
#2
Step 3.7 Flash
Step 3.7 Flash
206 t/s
#3
NVIDIA Nemotron Nano 9B V2
NVIDIA Nemotron Nano 9B V2
203 t/s
#4
NVIDIA Nemotron Nano 9B V2
NVIDIA Nemotron Nano 9B V2
197 t/s
#5
NVIDIA Nemotron 3 Nano
NVIDIA Nemotron 3 Nano
195 t/s

Velocidad de salida

93 modelos en total

Menor precio

#1
Llama 3.1 8B (Turbo, FP8)
Llama 3.1 8B (Turbo, FP8)
$0.02
#2
Llama 3.1 8B
Llama 3.1 8B
$0.02
#3
Gemma 4 E4B
Gemma 4 E4B
$0.03
#4
Gemma 4 E4B
Gemma 4 E4B
$0.03
#5
gpt-oss-20b (high)
gpt-oss-20b (high)
$0.04

Precio combinado (por 1M de tokens)

93 modelos en total

Indica un modelo de razonamiento

DeepInfra ofrece 93 modelos, cada uno con distintas características de inteligencia, rendimiento y precio. A continuación se comparan las métricas clave entre modelos.

  • En inteligencia, los mejores modelos en DeepInfra son GLM-5.2 (max) (FP4) (51), MiniMax-M3 (44) y DeepSeek V4 Pro (max) (FP4) (44).
  • En velocidad de salida, los modelos más rápidos son gpt-oss-120b (high) (Turbo) (253 t/s), Step 3.7 Flash (206 t/s) y NVIDIA Nemotron Nano 9B V2 (203 t/s).
  • En latencia, NVIDIA Nemotron 3 Nano (0.41s), Qwen3.5 35B A3B FP8 (0.52s) y Qwen3 Coder 480B (Turbo, FP4) (0.54s) ofrecen el menor tiempo hasta el primer token de respuesta.
  • En precios, Llama 3.1 8B (Turbo, FP8) ($0.02), Llama 3.1 8B ($0.02) y Gemma 4 E4B ($0.03) ofrecen los menores precios combinados por 1M de tokens.
  • En tamaño de ventana de contexto, GLM-5.2 (max) (FP4) (1M), DeepSeek V4 Pro (max) (FP4) (1M) y DeepSeek V4 Flash (max) (FP4) (1M) admiten las ventanas de contexto más grandes en DeepInfra.

Aspectos destacados

Artificial Analysis Intelligence Index · Higher is better
Output tokens per second · Higher is better
USD per 1M tokens (blended) · Lower is better

Evaluaciones de inteligencia

Artificial Analysis Intelligence Index

Artificial Analysis Intelligence Index v4.1 incorporates 9 evaluations: GDPval-AA v2, 𝜏³-Banking, Terminal-Bench v2.1, SciCode, Humanity's Last Exam, GPQA Diamond, CritPt, AA-Omniscience, AA-LCR
Estimate (independent evaluation forthcoming)
Reasoning models are indicated by a lightbulb icon

Artificial Analysis Intelligence Index v4.1 includes: GDPval-AA v2, 𝜏³-Banking, Terminal-Bench v2.1, SciCode, Humanity's Last Exam, GPQA Diamond, CritPt, AA-Omniscience, AA-LCR. See Intelligence Index methodology for further details, including a breakdown of each evaluation and how we run them.

Intelligence Evaluations

Intelligence evaluations measured independently by Artificial Analysis · Higher is better

Agentic real-world work tasks, (Elo-500)/2000

Agentic tool use

Agentic coding & terminal use

Coding

Reasoning & knowledge

Scientific reasoning

Physics reasoning

Long context reasoning

Agentic knowledge work, Elo

Agentic SaaS workflows

Legal agentic work, criterion pass rate

Agentic business operations

Instruction following

Long-horizon agentic tasks

Kubernetes incident root-cause analysis

Visual reasoning

Reasoning models are indicated by a lightbulb icon

While model intelligence generally translates across use cases, specific evaluations may be more relevant for certain use cases.

Artificial Analysis Intelligence Index v4.1 includes: GDPval-AA v2, 𝜏³-Banking, Terminal-Bench v2.1, SciCode, Humanity's Last Exam, GPQA Diamond, CritPt, AA-Omniscience, AA-LCR. See Intelligence Index methodology for further details, including a breakdown of each evaluation and how we run them.

Intelligence Index vs. Price

Blended at 7:2:1 (cache-input-output) · USD per 1M tokens (blended)
Most attractive quadrant
Reasoning models are indicated by a lightbulb icon

While higher intelligence models are typically more expensive, they do not all follow the same price-quality curve.

Artificial Analysis Intelligence Index v4.1 includes: GDPval-AA v2, 𝜏³-Banking, Terminal-Bench v2.1, SciCode, Humanity's Last Exam, GPQA Diamond, CritPt, AA-Omniscience, AA-LCR. See Intelligence Index methodology for further details, including a breakdown of each evaluation and how we run them.

Price per token, shown in USD per million tokens. Price is a blend of cache hit, input, and output token prices using the selected ratio (default 7:2:1 cache-input-output).

The blended cache price shown here uses cache hit price only. Other caching costs differ by provider:

  • Anthropic: charges a separate cache write fee, with different rates for 5-minute and 1-hour TTLs (1-hour TTL is more expensive).
  • Google (Vertex/Gemini): charges a per-hour cache storage fee in addition to cache hit pricing. Some providers also use tiered pricing for prompts above 200K tokens.
  • OpenAI, DeepSeek, others: typically charge only cache hit pricing with no write or storage fee.

See Prompt Caching for the full breakdown.

Figures represent performance of the model's first-party API (e.g. OpenAI for o1) or the median across providers where a first-party API is not available (e.g. Meta's Llama models).

Ventana de contexto

Context Window

Context window: tokens limit · Higher is better
Reasoning models are indicated by a lightbulb icon

Larger context windows are relevant to RAG (Retrieval Augmented Generation) LLM workflows which typically involve reasoning and information retrieval of large amounts of data.

Maximum number of combined input & output tokens. Output tokens commonly have a significantly lower limit (varied by model).

Precios

Intelligence Index vs. Price

Blended at 7:2:1 (cache-input-output) · USD per 1M tokens (blended)
Most attractive quadrant
Reasoning models are indicated by a lightbulb icon

While higher intelligence models are typically more expensive, they do not all follow the same price-quality curve.

Artificial Analysis Intelligence Index v4.1 includes: GDPval-AA v2, 𝜏³-Banking, Terminal-Bench v2.1, SciCode, Humanity's Last Exam, GPQA Diamond, CritPt, AA-Omniscience, AA-LCR. See Intelligence Index methodology for further details, including a breakdown of each evaluation and how we run them.

Price per token, shown in USD per million tokens. Price is a blend of cache hit, input, and output token prices using the selected ratio (default 7:2:1 cache-input-output).

The blended cache price shown here uses cache hit price only. Other caching costs differ by provider:

  • Anthropic: charges a separate cache write fee, with different rates for 5-minute and 1-hour TTLs (1-hour TTL is more expensive).
  • Google (Vertex/Gemini): charges a per-hour cache storage fee in addition to cache hit pricing. Some providers also use tiered pricing for prompts above 200K tokens.
  • OpenAI, DeepSeek, others: typically charge only cache hit pricing with no write or storage fee.

See Prompt Caching for the full breakdown.

Figures represent performance of the model's first-party API (e.g. OpenAI for o1) or the median across providers where a first-party API is not available (e.g. Meta's Llama models).

Resumen de rendimiento

Output Speed vs. Price

Output speed: output tokens per second · USD per 1M tokens (blended)
Most attractive quadrant
Reasoning models are indicated by a lightbulb icon

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).

Price per token, shown in USD per million tokens. Price is a blend of cache hit, input, and output token prices using the selected ratio (default 7:2:1 cache-input-output).

Velocidad

Medida por la velocidad de salida (tokens por segundo)

Output Speed

Output tokens per second · Higher is better
Reasoning models are indicated by a lightbulb icon

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 (e.g. OpenAI for o1) or the median across providers where a first-party API is not available (e.g. Meta's Llama models).

Latencia

Medida por el tiempo (segundos) hasta el primer token

Latency: Time To First Answer Token

Seconds to first answer token received · Accounts for reasoning model 'thinking' time
Reasoning models are indicated by a lightbulb icon

Time to first answer token received, in seconds, after API request sent. For reasoning models, this includes the 'thinking' time of the model before providing an answer. For models which do not support streaming, this represents time to receive the completion.

Tiempo de respuesta de extremo a extremo

Seconds to output 500 tokens, calculated based on time to first token, 'thinking' time for reasoning models, and output speed

End-to-End Response Time vs. Price

End-to-end response time: end-to-end seconds to output 500 tokens · USD per 1M tokens (blended)
Most attractive quadrant
Reasoning models are indicated by a lightbulb icon

Price per token, shown in USD per million tokens. Price is a blend of cache hit, input, and output token prices using the selected ratio (default 7:2:1 cache-input-output).

Seconds to receive a 500 token response. Key components:

  • Input time: Time to receive the first response token
  • Thinking time (only for reasoning models): Time reasoning models spend outputting tokens to reason prior to providing an answer. Amount of tokens based on the average reasoning tokens across a diverse set of 60 prompts (methodology details).
  • Answer time: Time to generate 500 output tokens, based on output speed

Figures represent median (P50) measurement over the past 72 hours to reflect sustained changes in performance.

Análisis adicional
Logo de Z AI
GLM-5.2 (max) (FP4)
1.05M
Abierto
51
--
56
1.02
45.62
35.68
Logo de MiniMax
MiniMax-M3
524k
Abierto
44
$0.12
31
1.18
81.12
63.95
Logo de DeepSeek
DeepSeek V4 Pro (max) (FP4)
1.05M
Abierto
44
$0.27
55
1.27
90.17
79.78
Logo de Kimi
Kimi K2.6 (FP4)
262k
Abierto
44
$0.25
61
0.86
81.63
72.62
Logo de DeepSeek
DeepSeek V4 Pro (high) (FP4)
65.5k
Abierto
43
$0.27
63
1.21
41.01
31.81
Logo de Xiaomi
MiMo-V2.5-Pro
65.5k
Abierto
42
$0.22
94
0.75
27.46
21.37
Logo de Kimi
Kimi K2.7 Code
262k
Abierto
42
$0.16
36
0.87
76.12
61.46
Logo de Tencent
Hy3 (FP8)
262k
Abierto
41
$0.03
48
0.63
52.93
41.84
Logo de Thinking Machines
Inkling (FP8)
131k
Abierto
41
$0.40
9
1.60
270.21
214.89
Logo de DeepSeek
DeepSeek V4 Flash (max) (FP4)
1.05M
Abierto
40
$0.04
34
1.19
179.58
163.81
Logo de Z AI
GLM-5.1 (FP4)
203k
Abierto
40
$0.17
52
0.97
83.66
73.05
Logo de Z AI
GLM-5 (FP4)
203k
Abierto
40*
--
44
0.97
82.94
70.60
Logo de NVIDIA
Nemotron 3 Ultra
262k
Abierto
38
--
90
4.57
35.39
25.27
Logo de NVIDIA
Nemotron 3 Ultra BF16
262k
Abierto
38
--
--
--
--
--
Logo de DeepSeek
DeepSeek V4 Flash (high) (FP4)
1.05M
Abierto
37
$0.03
34
1.08
51.63
36.02
Logo de Xiaomi
MiMo-V2.5
262k
Abierto
37
--
34
1.87
75.48
58.88
Logo de Alibaba
Qwen3.6 27B FP8
262k
Abierto
37
$0.18
49
1.30
127.89
116.35
Logo de Kimi
Kimi K2.5
262k
Abierto
35
$0.07
51
0.68
68.18
57.77
Logo de Z AI
GLM-5.1 (FP4)
203k
Abierto
35*
--
52
1.07
10.71
--
Logo de Kimi
Kimi K2.6 (FP4)
262k
Abierto
35*
--
54
0.85
10.15
--
Logo de Alibaba
Qwen3.5 27B (FP8)
262k
Abierto
34*
--
50
1.26
50.98
39.78
Logo de Z AI
GLM-4.7 (FP4)
203k
Abierto
34
$0.09
42
1.16
60.08
47.14
Logo de Alibaba
Qwen3.5 397B A17B (FP8)
262k
Abierto
34
$0.18
33
0.94
114.10
97.81
Logo de MiniMax
MiniMax-M2.5 (FP8)
197k
Abierto
34*
--
35
0.83
71.26
56.35
Logo de Z AI
GLM-5 (FP8)
203k
Abierto
32*
--
36
1.12
14.87
--
Logo de Alibaba
Qwen3.5 122B A10B (FP4)
262k
Abierto
32
$0.11
93
5.80
32.54
21.39
Logo de DeepSeek
DeepSeek V3.2 (FP4)
164k
Abierto
32
$0.06
16
0.69
154.42
122.98
Logo de Alibaba
Qwen3.5 397B A17B (FP8)
262k
Abierto
32*
--
34
0.94
15.85
--
Logo de Alibaba
Qwen3.6 35B A3B (FP8)
262k
Abierto
32
$0.11
25
1.17
239.53
218.14
Logo de Alibaba
Qwen3.6 27B FP8
262k
Abierto
30
$0.21
50
1.16
11.16
--
Logo de StepFun
Step 3.7 Flash
256k
Abierto
30
--
206
0.44
12.60
9.73
Logo de Google
Gemma 4 31B
262k
Abierto
29
$0.01
22
8.88
112.36
80.34
Logo de Alibaba
Qwen3.5 27B FP8
262k
Abierto
29*
--
52
1.13
10.72
--
Logo de Alibaba
Qwen3.5 35B A3B (FP8)
262k
Abierto
29*
--
84
0.55
30.32
23.81
Logo de Z AI
GLM-4.6 (FP4)
203k
Abierto
29
--
33
1.05
77.62
61.25
Logo de Xiaomi
MiMo-V2.5-Pro
65.5k
Abierto
28*
--
102
0.72
5.61
--
Logo de Alibaba
Qwen3.5 122B A10B (FP4)
262k
Abierto
28
$0.09
90
5.83
11.40
--
Logo de Z AI
GLM-4.7 (FP4)
203k
Abierto
27*
--
38
0.94
14.23
--
Logo de Google
Gemma 4 26B A4B
262k
Abierto
26
$0.02
15
1.22
172.40
136.94
Logo de NVIDIA
NVIDIA Nemotron 3 Super
262k
Abierto
25
$0.09
--
--
--
--
Logo de DeepSeek
DeepSeek V3.2
164k
Abierto
25*
--
19
1.51
28.43
--
Logo de Alibaba
Qwen3.6 35B A3B (FP8)
262k
Abierto
24
$0.30
23
2.55
24.36
--
Logo de Alibaba
Qwen3.5 35B A3B FP8
262k
Abierto
24
$0.07
83
0.52
6.55
--
Logo de OpenAI
gpt-oss-120b (high)
131k
Abierto
24
$0.02
42
0.70
60.20
47.60
Logo de OpenAI
gpt-oss-120b (high) (Turbo)
131k
Abierto
24
$0.06
253
0.70
10.57
7.89
Logo de Z AI
GLM-4.7-Flash
203k
Abierto
23*
--
49
1.44
52.76
41.05
Logo de Google
Gemma 4 31B (FP8)
262k
Abierto
22
$0.03
5
4.69
108.74
--
Logo de DeepSeek
DeepSeek V3.1 Terminus (FP4)
164k
Abierto
21*
--
47
0.58
11.21
--
Logo de DeepSeek
DeepSeek V3.1 (FP4)
164k
Abierto
21*
--
18
1.02
28.14
--
Logo de Google
Gemma 4 26B A4B (FP8)
262k
Abierto
20*
--
15
0.83
34.52
--
Logo de Alibaba
Qwen3.5 4B (FP8)
262k
Abierto
20*
--
23
0.72
110.38
87.73
Logo de DeepSeek
DeepSeek R1 0528
164k
Abierto
20*
--
29
0.83
87.19
69.09
Logo de Alibaba
Qwen3 235B A22B 2507 (FP8)
262k
Abierto
20
--
41
1.04
62.69
49.32
Logo de Alibaba
Qwen3 235B 2507 (FP8)
262k
Abierto
18*
--
13
0.76
40.09
--
Logo de Alibaba
Qwen3 Coder 480B (Turbo, FP4)
262k
Abierto
18*
--
68
0.54
7.91
--
Logo de Alibaba
Qwen3.5 4B FP8
262k
Abierto
16*
--
17
0.87
29.61
--
Logo de DeepSeek
DeepSeek V3 0324 (FP4)
164k
Abierto
15
$0.03
33
2.12
17.29
--
Logo de OpenAI
gpt-oss-20b (high)
131k
Abierto
15
$0.01
103
0.48
24.73
19.40
Logo de Mistral
Mistral Small 3.1
128k
Abierto
15
$0.02
28
1.09
18.68
--
Logo de Meta
Llama 4 Maverick (FP8)
1.05M
Abierto
14
$0.02
32
0.65
16.11
--
Logo de NVIDIA
NVIDIA Nemotron 3 Nano
262k
Abierto
14
$0.02
195
4.34
17.17
10.26
Logo de DeepSeek
DeepSeek V3 (Dec)
164k
Abierto
14
$0.02
19
0.88
26.85
--
Logo de Alibaba
Qwen3 Next 80B A3B
262k
Abierto
14*
--
120
0.70
4.86
--
Logo de NVIDIA
Llama Nemotron Super 49B v1.5
131k
Abierto
12*
--
80
4.56
35.94
25.11
Logo de Google
Gemma 4 E4B
262k
Abierto
12
--
90
0.80
28.57
22.22
Logo de Alibaba
Qwen3 32B (FP8)
41k
Abierto
12
--
53
0.63
47.71
37.67
Logo de Mistral
Mistral Small 3.2 (FP8)
128k
Abierto
11
$0.09
28
0.83
18.58
--
Logo de Alibaba
Qwen3 14B (FP8)
32.8k
Abierto
10
--
46
0.80
55.34
43.63
Logo de Meta
Llama 4 Scout
328k
Abierto
10
$0.0049
49
0.59
10.80
--
Logo de DeepSeek
DeepSeek R1 Distill Llama 70B
131k
Abierto
10*
--
30
0.85
85.56
67.76
Logo de Alibaba
Qwen2.5 72B
32.8k
Abierto
10*
--
25
3.19
23.51
--
Logo de Meta
Llama 3.3 70B (Turbo, FP8)
131k
Abierto
9
$0.01
9
2.34
60.04
--
Logo de Alibaba
Qwen3 30B (FP8)
41k
Abierto
9*
--
69
0.57
36.88
29.05
Logo de NVIDIA
NVIDIA Nemotron Nano 12B v2 VL (FP8)
131k
Abierto
9*
--
83
7.97
38.27
24.24
Logo de Google
Gemma 4 E4B
262k
Abierto
9*
--
92
0.77
6.22
--
Logo de NVIDIA
NVIDIA Nemotron Nano 9B V2
131k
Abierto
9*
--
197
3.50
16.18
10.15
Logo de NVIDIA
Llama Nemotron Super 49B v1.5
131k
Abierto
9*
--
80
5.71
11.95
--
Logo de Alibaba
Qwen3 32B (FP8)
41k
Abierto
9*
--
55
0.80
9.87
--
Logo de NVIDIA
Llama 3.1 Nemotron 70B
131k
Abierto
8*
--
76
6.42
13.00
--
Logo de Meta
Llama 3.1 8B (Turbo, FP8)
131k
Abierto
8
--
17
1.07
30.64
--
Logo de Meta
Llama 3.1 8B
131k
Abierto
8
--
17
1.04
30.52
--
Logo de Google
Gemma 3 27B
131k
Abierto
7
$0.07
17
1.30
30.41
--
Logo de NVIDIA
NVIDIA Nemotron 3 Nano
262k
Abierto
7*
--
194
0.41
2.99
--
Logo de NVIDIA
NVIDIA Nemotron Nano 9B V2
131k
Abierto
7*
--
203
3.35
5.82
--
Logo de Alibaba
Qwen3 14B (FP8)
41k
Abierto
7*
--
49
0.79
11.05
--
Logo de Mistral
Mistral Small 3
32.8k
Abierto
7*
--
47
0.80
11.47
--
Logo de Alibaba
Qwen3 30B (FP8)
41k
Abierto
7*
--
60
0.55
8.87
--
Logo de Meta
Llama 3.1 70B
131k
Abierto
7*
--
37
1.65
15.08
--
Logo de Meta
Llama 3.1 70B (Turbo, FP8)
131k
Abierto
7*
--
34
1.65
16.55
--
Logo de Google
Gemma 3 12B
131k
Abierto
6
$0.07
50
0.78
10.73
--
Logo de Nous Research
Hermes 3 - Llama-3.1 70B
131k
Abierto
5*
--
35
1.98
16.19
--
Logo de Microsoft
Phi-4
16.4k
Abierto
5*
--
73
0.85
7.72
--
Logo de NVIDIA
NVIDIA Nemotron Nano 12B v2 VL (FP8)
131k
Abierto
5*
--
96
5.48
10.69
--
Logo de Meta
Llama 3.2 11B (Vision)
131k
Abierto
3*
--
5
4.15
102.96
--
Logo de Meta
Llama 3 8B
8.19k
Abierto
1*
--
--
--
--
--
Logo de Google
Gemma 3 4B
131k
Abierto
1*
--
34
0.89
15.51
--

Definiciones clave

Maximum number of combined input & output tokens. Output tokens commonly have a significantly lower limit (varied by model).

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).

Time to first token received, in seconds, after API request sent. For reasoning models which share reasoning tokens, this will be the first reasoning token. For models which do not support streaming, this represents time to receive the completion.

Average cost per task in the index. Costs are split by input, cache hit, cache write, reasoning, and answer token pricing where canonical token counts are available.

Price per token included in the request/message sent to the API, represented as USD per million Tokens.

Price per token for cached prompts (previously processed), typically offering a significant discount compared to regular input price, represented as USD per million tokens. The values shown here are the cache hit price; cache write and cache storage are billed separately and vary by provider — see "Cache pricing by provider" for detail.

Price per token to write prompt tokens into the cache so that later requests can hit them, represented as USD per million tokens. Some providers charge a premium over the standard input price to create a cache entry (e.g. Anthropic), while others cache automatically with no separate write fee.

Price per token generated by the model (received from the API), represented as USD per million Tokens.

Metrics are 'live' and are based on the past 72 hours of measurements, measurements are taken 8 times a day for single requests and 2 times per day for parallel requests.

Preguntas frecuentes

Preguntas comunes sobre DeepInfra

DeepInfra ofrece 93 modelos que monitoreamos: GLM-5.2 (max) (FP4), MiniMax-M3, DeepSeek V4 Pro (max) (FP4), Kimi K2.6 (FP4), DeepSeek V4 Pro (high) (FP4), MiMo-V2.5-Pro, Kimi K2.7 Code, Hy3 (FP8), Inkling (FP8), DeepSeek V4 Flash (max) (FP4), GLM-5.1 (FP4), GLM-5 (FP4), Nemotron 3 Ultra, DeepSeek V4 Flash (high) (FP4), MiMo-V2.5, Qwen3.6 27B FP8, Kimi K2.5, GLM-5.1 (FP4), Kimi K2.6 (FP4), Qwen3.5 27B (FP8), GLM-4.7 (FP4), Qwen3.5 397B A17B (FP8), MiniMax-M2.5 (FP8), GLM-5 (FP8), Qwen3.5 122B A10B (FP4), DeepSeek V3.2 (FP4), Qwen3.5 397B A17B (FP8), Qwen3.6 35B A3B (FP8), Qwen3.6 27B FP8, Step 3.7 Flash, Gemma 4 31B, Qwen3.5 27B FP8, Qwen3.5 35B A3B (FP8), GLM-4.6 (FP4), MiMo-V2.5-Pro, Qwen3.5 122B A10B (FP4), GLM-4.7 (FP4), Gemma 4 26B A4B, DeepSeek V3.2, Qwen3.6 35B A3B (FP8), Qwen3.5 35B A3B FP8, gpt-oss-120b (high), gpt-oss-120b (high) (Turbo), GLM-4.7-Flash, Gemma 4 31B (FP8), DeepSeek V3.1 Terminus (FP4), DeepSeek V3.1 (FP4), Gemma 4 26B A4B (FP8), Qwen3.5 4B (FP8), DeepSeek R1 0528, Qwen3 235B A22B 2507 (FP8), Qwen3 235B 2507 (FP8), Qwen3 Coder 480B (Turbo, FP4), Qwen3.5 4B FP8, DeepSeek V3 0324 (FP4), gpt-oss-20b (high), Mistral Small 3.1, Llama 4 Maverick (FP8), NVIDIA Nemotron 3 Nano, DeepSeek V3 (Dec), Qwen3 Next 80B A3B, Llama Nemotron Super 49B v1.5, Gemma 4 E4B, Qwen3 32B (FP8), Mistral Small 3.2 (FP8), Qwen3 14B (FP8), Llama 4 Scout, DeepSeek R1 Distill Llama 70B, Qwen2.5 72B, Llama 3.3 70B (Turbo, FP8), Qwen3 30B (FP8), NVIDIA Nemotron Nano 12B v2 VL (FP8), Gemma 4 E4B, NVIDIA Nemotron Nano 9B V2, Llama Nemotron Super 49B v1.5, Qwen3 32B (FP8), Llama 3.1 Nemotron 70B, Llama 3.1 8B (Turbo, FP8), Llama 3.1 8B, Gemma 3 27B, NVIDIA Nemotron 3 Nano, NVIDIA Nemotron Nano 9B V2, Qwen3 14B (FP8), Mistral Small 3, Qwen3 30B (FP8), Llama 3.1 70B, Llama 3.1 70B (Turbo, FP8), Gemma 3 12B, Hermes 3 - Llama-3.1 70B, Phi-4, NVIDIA Nemotron Nano 12B v2 VL (FP8), Llama 3.2 11B (Vision) y Gemma 3 4B.

El modelo más inteligente disponible en DeepInfra es GLM-5.2 (max) (FP4), con una puntuación de 51 en el Índice de Inteligencia.

El modelo más rápido en DeepInfra por velocidad de salida es gpt-oss-120b (high) (Turbo), con 253.4 tokens por segundo.

El modelo con el menor tiempo hasta el primer token de respuesta en DeepInfra es NVIDIA Nemotron 3 Nano, con 0.41s. Una menor latencia significa una respuesta inicial más rápida.

El modelo más económico en DeepInfra por precio combinado es Llama 3.1 8B (Turbo, FP8), a $0.02 por 1M de tokens (proporción 7:2:1 de aciertos de caché/entrada/salida).

Los precios en DeepInfra varían hasta 55x entre modelos, desde $0.02 por 1M de tokens para Llama 3.1 8B (Turbo, FP8) hasta $1.20 por 1M de tokens para Llama 3.1 Nemotron 70B.

Sí, DeepInfra ofrece una API compatible con OpenAI, lo que facilita cambiar desde OpenAI o usar integraciones existentes del SDK de OpenAI.

89 de 93 modelos en DeepInfra admiten el modo JSON para salida estructurada.

90 de 93 modelos en DeepInfra admiten llamadas a funciones (uso de herramientas).

Sí, los 93 modelos en DeepInfra son modelos de pesos abiertos.

Sí, el rendimiento de un proveedor puede variar con el tiempo debido a cambios de infraestructura, balanceo de carga y actualizaciones. Medimos continuamente a todos los proveedores y mostramos las tendencias históricas de rendimiento en los gráficos "a lo largo del tiempo".

Al elegir un modelo en DeepInfra, considera: inteligencia (para tareas sensibles a la calidad), velocidad de salida (para tareas intensivas en rendimiento), latencia (para aplicaciones interactivas que requieren primeras respuestas rápidas), precios (para cargas de trabajo sensibles al costo) y funcionalidades como el tamaño de la ventana de contexto, el modo JSON o las llamadas a funciones.