Multilingual AI Model Benchmark Compare Leading LLMs by Language

Explore how leading large language models (LLMs) perform across multiple languages on Artificial Analysis' Multilingual Index, including the Global-MMLU-Lite benchmark. Filter by language and model, view trade-offs between accuracy, speed, and cost, and find the best LLM for your multilingual use case.

For details on datasets and methodology, see the FAQ page.

🌐All (average)

Top Models
#1
Gemini 3.1 Pro PreviewGemini 3.1 Pro Preview
93#2
Gemini 3 Pro Preview (high)Gemini 3 Pro Preview (high)
92#3
Claude Opus 4.6 (Adaptive Reasoning, Max Effort)Claude Opus 4.6 (max)
92#4
Gemini 3 Flash Preview (Reasoning)Gemini 3 Flash
91#5
Claude Opus 4.5 (Reasoning)Claude Opus 4.5
91

🇬🇧English

Top Models
#1
Claude Opus 4.6 (Adaptive Reasoning, Max Effort)Claude Opus 4.6 (max)
95#2
Gemini 3.1 Pro PreviewGemini 3.1 Pro Preview
95#3
Gemini 3 Flash Preview (Reasoning)Gemini 3 Flash
95#4
Claude Opus 4.5 (Non-reasoning)Claude Opus 4.5
94#5
GPT-5.1 (high)GPT-5.1 (high)
94

🇨🇳Chinese

Top Models
#1
Claude Opus 4.6 (Adaptive Reasoning, Max Effort)Claude Opus 4.6 (max)
94#2
Gemini 3.1 Pro PreviewGemini 3.1 Pro Preview
94#3
Gemini 3 Pro Preview (high)Gemini 3 Pro Preview (high)
94#4
Gemini 3 Flash Preview (Reasoning)Gemini 3 Flash
93#5
Claude Opus 4.5 (Reasoning)Claude Opus 4.5
92

🇮🇳Hindi

Top Models
#1
Gemini 3.1 Pro PreviewGemini 3.1 Pro Preview
94#2
Claude Opus 4.6 (Adaptive Reasoning, Max Effort)Claude Opus 4.6 (max)
92#3
Gemini 3 Pro Preview (high)Gemini 3 Pro Preview (high)
91#4
GPT-5 (high)GPT-5 (high)
91#5
Claude Opus 4.5 (Non-reasoning)Claude Opus 4.5
91

🇪🇸Spanish

Top Models
#1
Gemini 3.1 Pro PreviewGemini 3.1 Pro Preview
94#2
Gemini 3 Pro Preview (high)Gemini 3 Pro Preview (high)
94#3
Gemini 3 Flash Preview (Reasoning)Gemini 3 Flash
94#4
Claude Opus 4.6 (Adaptive Reasoning, Max Effort)Claude Opus 4.6 (max)
94#5
Claude Opus 4.5 (Non-reasoning)Claude Opus 4.5
94

🇫🇷French

Top Models
#1
Gemini 3.1 Pro PreviewGemini 3.1 Pro Preview
94#2
GPT-5.1 (high)GPT-5.1 (high)
94#3
Gemini 3 Pro Preview (high)Gemini 3 Pro Preview (high)
93#4
Claude Opus 4.6 (Adaptive Reasoning, Max Effort)Claude Opus 4.6 (max)
93#5
GPT-5 (high)GPT-5 (high)
93

Other Countries

🇸🇦 Arabic
Gemini 3.1 Pro PreviewGemini 3.1 Pro Preview
🇧🇩 Bengali
Gemini 3.1 Pro PreviewGemini 3.1 Pro Preview
🇵🇹 Portuguese
Gemini 3.1 Pro PreviewGemini 3.1 Pro Preview
🇮🇩 Indonesian
Gemini 3.1 Pro PreviewGemini 3.1 Pro Preview
🇯🇵 Japanese
Gemini 3.1 Pro PreviewGemini 3.1 Pro Preview
🇰🇪 Swahili
Gemini 3.1 Pro PreviewGemini 3.1 Pro Preview
🇩🇪 German
Gemini 3.1 Pro PreviewGemini 3.1 Pro Preview
🇰🇷 Korean
Claude Opus 4.6 (Adaptive Reasoning, Max Effort)Claude Opus 4.6 (max)
🇮🇹 Italian
Gemini 3.1 Pro PreviewGemini 3.1 Pro Preview
🇳🇬 Yoruba
Gemini 3.1 Pro PreviewGemini 3.1 Pro Preview
🇲🇲 Burmese
Gemini 3.1 Pro PreviewGemini 3.1 Pro Preview

Overview

Artificial Analysis Multilingual Index

Higher is better
All (average)
Chinese
English
Hindi
Spanish
Reasoning models are indicated by a lightbulb icon.

An index assessing multilingual performance in general reasoning across multiple languages. Results are computed across English, Chinese, Hindi, Spanish, French, Arabic, Bengali, Portuguese, Indonesian, Japanese, Swahili, German, Korean, Italian, Yoruba, Burmese. See Multilingual Intelligence Index methodology for further details.

Multilingual Index Across Languages (Normalized)

Scores are normalized per language across all models tested, where green represents the highest score for that language and red represents the lowest score for that language.
Reasoning models are indicated by a lightbulb icon

An index assessing multilingual performance in general reasoning across multiple languages. Results are computed across English, Chinese, Hindi, Spanish, French, Arabic, Bengali, Portuguese, Indonesian, Japanese, Swahili, German, Korean, Italian, Yoruba, Burmese. See Multilingual Intelligence Index methodology for further details.

Multilingual Index

Multilingual Index: Average Across All Languages

Artificial Analysis Multilingual Index · Average across all languages · Higher is better
Reasoning models are indicated by a lightbulb icon.

An index assessing multilingual performance in general reasoning across multiple languages. Results are computed across English, Chinese, Hindi, Spanish, French, Arabic, Bengali, Portuguese, Indonesian, Japanese, Swahili, German, Korean, Italian, Yoruba, Burmese. See Multilingual Intelligence Index methodology for further details.

Multilingual Index: Average vs. Output Speed

Artificial Analysis Multilingual Index · 1,000 input tokens
Most attractive quadrant
Anthropic
Meta
MiniMax
Mistral
OpenAI
Reasoning models are indicated by a lightbulb icon.

There is a trade-off between model quality and output speed, with higher intelligence models typically having lower output speed.

An index assessing multilingual performance in general reasoning across multiple languages. Results are computed across English, Chinese, Hindi, Spanish, French, Arabic, Bengali, Portuguese, Indonesian, Japanese, Swahili, German, Korean, Italian, Yoruba, Burmese. See Multilingual Intelligence Index methodology for further details.

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).

Multilingual Index: Average vs. Price

Artificial Analysis Multilingual Index · Average across all languages
Most attractive quadrant
Anthropic
DeepSeek
Google
Meta
MiniMax
Mistral
OpenAI
xAI
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.

An index assessing multilingual performance in general reasoning across multiple languages. Results are computed across English, Chinese, Hindi, Spanish, French, Arabic, Bengali, Portuguese, Indonesian, Japanese, Swahili, German, Korean, Italian, Yoruba, Burmese. See Multilingual Intelligence Index methodology for further details.

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).

Global-MMLU-Lite

Multilingual Global-MMLU-Lite: Average

Average across all languages · Higher is better
Reasoning models are indicated by a lightbulb icon.

A multilingual version of Massive Multitask Language Understanding, evaluated across multiple languages. Tests general knowledge and reasoning ability in areas like science, humanities, mathematics and more. See methodology for further details.

Pricing

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). Blended price charts use Anthropic cache write price for the input leg.
  • 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).

Speed & Latency

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).

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.

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).