Inkling (xhigh) logo
Thinking Machines
•

Open weights model

•

Released July 2026

Análisis de inteligencia, rendimiento y precio de Inkling (xhigh)

Probar Benchmarks de proveedores de API

Resumen del modelo

Inteligencia

41
Índice de Inteligencia de Artificial Analysis
4 de 4 unidades para Inteligencia.

Velocidad

85.4
Tokens de salida por segundo
3 de 4 unidades para Velocidad.

Precio

Entrada
1,87 US$
por 1 M de tokens
Salida
4,68 US$
por 1 M de tokens
4 de 4 unidades para Precio.

Precio de acierto de caché

0,374 US$
USD por 1 M de tokens
4 de 4 unidades para Precio de acierto de caché.

Verbosidad

130M
Tokens de salida del Índice de Inteligencia
3 de 4 unidades para Verbosidad.

Inkling (xhigh) está entre los modelos líderes en inteligencia, pero tiene un precio especialmente elevado en comparación con otros modelos de pesos abiertos y tamaño similar. También es más rápido que el promedio; sin embargo, es algo verboso. El modelo admite entrada de texto, imágenes y voz, genera texto y tiene una ventana de contexto de 1M tokens.

Inkling (xhigh) obtiene 41 puntos en el Índice de Inteligencia de Artificial Analysis, lo que lo sitúa muy por encima del promedio entre modelos comparables (con una mediana de 25). Al evaluar el Índice de Inteligencia, generó 130M tokens, lo que es algo verboso frente a la mediana de 99M.

El precio de Inkling (xhigh) es de $1.87 por 1 M de tokens de entrada (elevado; mediana: $0.43) y $4.68 por 1 M de tokens de salida (elevado; mediana: $1.23).

Con 85 tokens por segundo, Inkling (xhigh) es más rápido que el promedio (61).

RazonamientoSí

Esta página muestra la versión con razonamiento de este modelo.

También puede existir una variante sin razonamiento.

Modalidad de entrada

Admite: texto, imágenes y voz

Modalidad de salida

Admite: texto

Ventana de contexto1M
~1500 páginas A4 con fuente Arial de 12 puntos
Parámetros totales975B
Parámetros activos41B
Número de parámetros activos por token durante la inferencia
LicenciaApache 2.0
Pesos del modeloHugging Face

Las métricas se comparan con modelos de la misma clase:

  • Modelos sin razonamiento → se comparan solo con otros modelos sin razonamiento
  • Modelos de razonamiento → se comparan tanto con modelos de razonamiento como sin razonamiento
  • Modelos de pesos abiertos → se comparan solo con otros modelos de pesos abiertos de la misma categoría de tamaño:
    • Muy pequeño: ≤4B parámetros
    • Pequeño: 4B–40B parámetros
    • Mediano: 40B–150B parámetros
    • Grande: >150B parámetros
  • Modelos propietarios → se comparan con modelos propietarios y de pesos abiertos del mismo rango de precio mediante una proporción combinada de precios de entrada/salida de 3:1:
    • <$0.15 por 1 M de tokens
    • $0.15–$1 por 1 M de tokens
    • >$1 por 1 M de tokens

Aspectos destacados

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

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
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.

Artificial Analysis Intelligence Index by Open Weights / Proprietary

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
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.

Indicates whether the model weights are available. Models are labelled as 'Commercial Use Restricted' if the weights are available but commercial use is limited (typically requires obtaining a paid license).

Benchmarks

Intelligence Evaluations

Intelligence evaluations measured independently by Artificial Analysis · Higher is better
GDPval-AA v2

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

𝜏³-Banking

Agentic tool use

Terminal-Bench v2.1

Agentic coding & terminal use

SciCode

Coding

Humanity's Last Exam

Reasoning & knowledge

GPQA Diamond

Scientific reasoning

CritPt

Physics reasoning

AA-Omniscience Accuracy

Knowledge

AA-Omniscience Non-Hallucination Rate

1 - hallucination rate

AA-LCR

Long context reasoning

AA-Briefcase

Agentic knowledge work, Elo

AutomationBench-AA

Agentic SaaS workflows

Harvey LAB-AA

Legal agentic work, criterion pass rate

EnterpriseOps-Gym-AANew

Agentic business operations

IFBench

Instruction following

APEX-Agents-AA

Long-horizon agentic tasks

ITBench-AA

Kubernetes incident root-cause analysis

MMMU-Pro

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.

AA-Briefcase

AA-Briefcase Elo

AA-Briefcase is an agentic knowledge work benchmark developed by Artificial Analysis. AA-Briefcase Elo is a combined metric that aggregates rubric pass rate, analytical quality Elo and presentation Elo · Higher is better
Reasoning models are indicated by a lightbulb icon

AA-Briefcase Elo is a combined metric that aggregates analytical quality Elo, presentation Elo, and rubric pass rate, with rubric performance converted into Elo via synthetic head-to-head matches. Elo and 95% confidence interval bounds are clamped at 0.

AA-Omniscience

AA-Omniscience Index

AA-Omniscience Index (higher is better) measures knowledge reliability and hallucination. It rewards correct answers, penalizes hallucinations, and has no penalty for refusing to answer. Scores range from -100 to 100, where 0 means as many correct as incorrect answers, and negative scores mean more incorrect than correct.
Reasoning models are indicated by a lightbulb icon

AA-Omniscience Index (higher is better) measures knowledge reliability and hallucination. It rewards correct answers, penalizes hallucinations, and has no penalty for refusing to answer. Scores range from -100 to 100, where 0 means as many correct as incorrect answers, and negative scores mean more incorrect than correct.

Índice de apertura

Artificial Analysis Openness Index: Score

Openness Index assesses model openness on a 0 to 100 normalized scale (higher is more open)
Reasoning models are indicated by a lightbulb icon

Comparaciones del Índice de Inteligencia

Intelligence Index vs. Cost per Intelligence Index Task

Artificial Analysis Intelligence Index · Weighted average cost (USD) per Artificial Analysis Intelligence Index task
Most attractive quadrant
Reasoning models are indicated by a lightbulb icon

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.

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.

Uso de tokens

Output Tokens per Intelligence Index Task

Weighted average number of output tokens used to run one task in the Artificial Analysis Intelligence Index
Reasoning models are indicated by a lightbulb icon

The number of tokens required per Intelligence Index task. This is calculated by multiplying the output tokens per eval by the relative weights of each benchmark in the Intelligence Index, then dividing by task count (excluding repeats).

Costo

Cost per Intelligence Index Task

Weighted average cost (USD) per Artificial Analysis Intelligence Index task, segmented by token type. Lower is better
Reasoning models are indicated by a lightbulb icon

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.

Cost to Run Artificial Analysis Intelligence Index

Cost (USD) to run all evaluations in the Artificial Analysis Intelligence Index
Reasoning models are indicated by a lightbulb icon

The cost to run the evaluations in the Artificial Analysis Intelligence Index, calculated using the model's input, cache hit, cache write, reasoning, and answer token prices and the number of tokens used across evaluations (excluding repeats).

Pricing: Cache Hit, Input, and Output

Price (USD per M Tokens)
Reasoning models are indicated by a lightbulb icon

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

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.

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

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).

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).

Time per Intelligence Index Task

Weighted average decode time (minutes) per task; excludes TTFT and overhead time · Lower is better
Reasoning models are indicated by a lightbulb icon

The weighted average time (seconds) per Artificial Analysis Intelligence Index task. This is calculated by dividing output tokens per task by output speed, weighted by the relative weights of each benchmark in the Intelligence Index.

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

Seconds to output 500 tokens, including reasoning model 'thinking' time · Lower is better
Reasoning models are indicated by a lightbulb icon

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 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).

Tamaño del modelo (solo modelos de pesos abiertos)

Model Size: Total and Active Parameters

Comparison between total model parameters and parameters active during inference
Reasoning models are indicated by a lightbulb icon

The total number of trainable weights and biases in the model, expressed in billions. These parameters are learned during training and determine the model's ability to process and generate responses.

The number of parameters actually executed during each inference forward pass, expressed in billions. For Mixture of Experts (MoE) models, a routing mechanism selects a subset of experts per token, resulting in fewer active than total parameters. Dense models use all parameters, so active equals total.

Preguntas frecuentes

Preguntas comunes sobre Inkling (xhigh)

Inkling (xhigh) se lanzó el 15 de julio de 2026.

Inkling (xhigh) fue creado por Thinking Machines.

Inkling (xhigh) obtiene 41 puntos en el Índice de Inteligencia de Artificial Analysis, lo que lo sitúa muy por encima del promedio entre otros modelos de pesos abiertos y tamaño similar (mediana: 25).

Inkling (xhigh) genera 85.4 tokens de salida por segundo (según la API de Thinking Machines), un valor por encima del promedio frente a otros modelos de pesos abiertos y tamaño similar (mediana: 60.6 t/s).

Inkling (xhigh) tiene un tiempo hasta el primer token (TTFT) de 1.81 s (según la API de Thinking Machines), un valor mejor que el promedio frente a otros modelos de pesos abiertos y tamaño similar (mediana: 1.85 s).

Inkling (xhigh) cuesta $1.87 por 1 M de tokens de entrada (en el extremo superior; mediana: $0.59) y $4.68 por 1 M de tokens de salida (en el extremo superior; mediana: $2.20), según la API de Thinking Machines.

Inkling (xhigh) cuesta $1.87 por 1 M de tokens de entrada y $4.68 por 1 M de tokens de salida (según la API de Thinking Machines). Con una tarifa combinada (proporción 7:2:1 entre aciertos de caché, entrada y salida), equivale a $1.10 por 1 M de tokens. El precio puede variar según el proveedor. Comparar precios de proveedores

Al evaluarlo en el Índice de Inteligencia, Inkling (xhigh) generó 130M tokens de salida, un valor algo mayor que el promedio frente a otros modelos de pesos abiertos y tamaño similar (mediana: 99M).

Sí, Inkling (xhigh) es un modelo de razonamiento. Usa pensamiento extendido o razonamiento en cadena para resolver problemas complejos antes de dar una respuesta.

Inkling (xhigh) admite entrada de texto, imágenes y voz.

Inkling (xhigh) admite salida de texto.

Sí, Inkling (xhigh) admite imágenes como entrada y puede analizarlas, describirlas y responder preguntas sobre ellas.

Sí, Inkling (xhigh) es multimodal. Puede procesar entradas de texto, imágenes y voz y generar salidas de texto.

Inkling (xhigh) tiene una ventana de contexto de 1.0M tokens. Esto determina cuánto texto e historial de conversación puede procesar el modelo en una sola solicitud.

Sí, Inkling (xhigh) es un modelo de pesos abiertos. Los pesos del modelo están disponibles públicamente y pueden descargarse para alojarlo por cuenta propia.

Inkling (xhigh) tiene 975 mil millones de parámetros (41 mil millones activos).

Inkling (xhigh) es un modelo de mezcla de expertos (MoE) con 975 mil millones de parámetros totales, pero durante la inferencia solo usa 41 mil millones de parámetros activos.

Inkling (xhigh) se publica bajo la licencia Apache 2.0, que permite el uso comercial. Ver licencia

Inkling (xhigh) obtiene 41 puntos en el Índice de Inteligencia de Artificial Analysis. Este benchmark compuesto evalúa los modelos en razonamiento, conocimiento, matemáticas y programación.

Sí, Inkling (xhigh) está disponible mediante API a través de 3 proveedores. Comparar proveedores de API

Inkling (xhigh) está disponible a través de 3 proveedores de API. Comparar proveedores