Microsoft Azure: Models Quality, Performance & Price
Analysis of Microsoft Azure's models across key metrics including quality, price, output speed, latency, context window & more. This analysis is intended to support you in choosing the best model provided by Microsoft Azure for your use-case. For more details including relating to our methodology, see our FAQs. Models analyzed: GPT-4o (Aug '24), GPT-4o (May '24), GPT-4o mini, GPT-4o (Nov '24), Llama 3.3 70B, Llama 3.1 405B, Llama 3.1 70B, Llama 3.1 8B, Mistral Large 2 (Jul '24), Mistral Large 2 (Nov '24), Command-R+ (Apr '24), Command-R (Mar '24), Phi-3 Medium 14B, Jamba 1.5 Large, Jamba 1.5 Mini, GPT-4 Turbo, Llama 3 8B, Mistral Small (Feb '24), Mistral Large (Feb '24), and Jamba Instruct.
Link:
Azure Model Comparison Summary
Quality:
Mistral Large (Feb '24)Ā andĀ 
Llama 3 8BĀ are the highest quality models offered by Azure, followed by 
Command-R+ (Apr '24), Command-R (Mar '24) & 
Jamba Instruct.Output Speed (tokens/s):
GPT-4o mini (171 t/s)Ā andĀ Llama 3.1 8B (163 t/s)Ā are the fastest models offered by Azure, followed by GPT-4o (May '24), Jamba 1.5 Mini & Command-R (Mar '24).Latency (seconds): Llama 3.1 8B (0.31s)Ā and Ā Llama 3 8B (0.39s)Ā are the lowest latency models offered by Azure, followed by Mistral Small (Feb '24), Phi-3 Medium 14B & Command-R (Mar '24).Blended Price ($/M tokens): Jamba 1.5 Mini ($0.25)Ā andĀ GPT-4o mini ($0.26)Ā are the cheapest models offered by Azure, followed by Phi-3 Medium 14B, Llama 3.1 8B & Llama 3 8B.Context Window Size: Jamba 1.5 Large (256k)Ā andĀ Jamba 1.5 Mini (256k)Ā are the largest context window models offered by Azure, followed by Jamba Instruct, GPT-4o (Aug '24) & GPT-4o (May '24).
Mistral Large (Feb '24)Ā andĀ Llama 3 8BĀ are the highest quality models offered by Azure, followed by Command-R+ (Apr '24), Command-R (Mar '24) & Jamba Instruct.Output Speed (tokens/s): GPT-4o mini (171 t/s)Ā andĀ Llama 3.1 8B (163 t/s)Ā are the fastest models offered by Azure, followed by GPT-4o (May '24), Jamba 1.5 Mini & Command-R (Mar '24).Latency (seconds): Llama 3.1 8B (0.31s)Ā and Ā Llama 3 8B (0.39s)Ā are the lowest latency models offered by Azure, followed by Mistral Small (Feb '24), Phi-3 Medium 14B & Command-R (Mar '24).Blended Price ($/M tokens): Jamba 1.5 Mini ($0.25)Ā andĀ GPT-4o mini ($0.26)Ā are the cheapest models offered by Azure, followed by Phi-3 Medium 14B, Llama 3.1 8B & Llama 3 8B.Context Window Size: Jamba 1.5 Large (256k)Ā andĀ Jamba 1.5 Mini (256k)Ā are the largest context window models offered by Azure, followed by Jamba Instruct, GPT-4o (Aug '24) & GPT-4o (May '24).Highlights
Quality
Artificial Analysis Quality Index; Higher is better
Speed
Output Tokens per Second; Higher is better
Price
USD per 1M Tokens; Lower is better
Parallel Queries:
Prompt Length:
Quality vs. Output Speed, Price
Artificial Analysis Quality Index; Output Speed: Output Tokens per Second; Price: USD per 1M Tokens
Most attractive quadrant
Size represents Price (USD per M Tokens)
Artificial Analysis Quality Index: Average result across our evaluations covering different dimensions of model intelligence. Currently includes MMLU, GPQA, Math & HumanEval. OpenAI o1 model figures are preliminary and are based on figures stated by OpenAI. See methodology for more details.
Output Speed: 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: Price per token, represented as USD per million Tokens. Price is a blend of Input & Output token prices (3:1 ratio).
Function (Tool) Calling & JSON Mode
| Models | Function calling | JSON Mode |
|---|---|---|
| GPT-4o (Aug '24), Azure | ||
| GPT-4o (May '24), Azure | ||
| Llama 3.3 70B, Azure | ||
| Mistral Large 2 (Jul '24), Azure | ||
| Mistral Large 2 (Nov '24), Azure | ||
| Llama 3.1 405B, Azure | ||
| GPT-4o mini, Azure | ||
| GPT-4o (Nov '24), Azure | ||
| Llama 3.1 70B, Azure | ||
| Jamba 1.5 Large, Azure | ||
| Llama 3.1 8B, Azure | ||
| Command-R+ (Apr '24), Azure | ||
| Phi-3 Medium 14B, Azure |
Function (Tool) Calling: Indicates whether the provider supports function calling in their API. Function calling is also known as 'Tool Calling'.
JSON Mode: Indicates whether the provider supports JSON mode in their API. When JSON mode is enabled, the models will always return a valid JSON object.
Quality & Context Window
Quality Evaluations
Evaluation results measured independently by Artificial Analysis; Higher is better
Artificial Analysis Quality Index
Reasoning & Knowledge (MMLU)
Scientific Reasoning & Knowledge (GPQA Diamond)
Quantitative Reasoning (MATH-500)
Coding (HumanEval)
Artificial Analysis Multilingual Index
Artificial Analysis Quality Index: Average result across our evaluations covering different dimensions of model intelligence. Currently includes MMLU, GPQA, Math & HumanEval. OpenAI o1 model figures are preliminary and are based on figures stated by OpenAI. See methodology for more details.
Artificial Analysis Multilingual Index: Reflects performance across a range of languages. Calculated as the average of Multilingual MMLU (general reasoning) and MGSM (mathematical reasoning) evaluation scores. See our multilingual capabilities page for further analysis.
Quality vs. Context Window, Input Token Price
Artificial Analysis Quality Index; Context Window: Tokens Limit; Input Price: USD per 1M Input Tokens
Most attractive quadrant
Size represents Input Price (USD per M Input Tokens)
Artificial Analysis Quality Index: Average result across our evaluations covering different dimensions of model intelligence. Currently includes MMLU, GPQA, Math & HumanEval. OpenAI o1 model figures are preliminary and are based on figures stated by OpenAI. See methodology for more details.
Context window: Maximum number of combined input & output tokens. Output tokens commonly have a significantly lower limit (varied by model).
Input Price: Price per token included in the request/message sent to the API, represented as USD per million Tokens.
Context Window
Context Window: Tokens Limit; Higher is better
Context window: Maximum number of combined input & output tokens. Output tokens commonly have a significantly lower limit (varied by model).
Pricing
Quality vs. Price
Artificial Analysis Quality Index; Price: USD per 1M Tokens
Most attractive quadrant
Artificial Analysis Quality Index: Average result across our evaluations covering different dimensions of model intelligence. Currently includes MMLU, GPQA, Math & HumanEval. OpenAI o1 model figures are preliminary and are based on figures stated by OpenAI. See methodology for more details.
Price: Price per token, represented as USD per million Tokens. Price is a blend of Input & Output token prices (3:1 ratio).
Median across providers: Figures represent median (P50) across all providers which support the model.
Pricing: Input and Output Prices
Price: USD per 1M Tokens
Input price
Output price
Input Price: Price per token included in the request/message sent to the API, represented as USD per million Tokens.
Output Price: Price per token generated by the model (received from the API), represented as USD per million Tokens.
Median across providers: Figures represent median (P50) across all providers which support the model.
Pricing: Cached Inputs
Price: USD per 1M Tokens
Input (standard)
Cache write
Cache hit
Cache storage
Output (standard)
Input Price: Price per token included in the request/message sent to the API, represented as USD per million Tokens.
Output Price: Price per token generated by the model (received from the API), represented as USD per million Tokens.
Cache Write: One-time cost charged when storing a prompt in the cache for future reuse, represented as USD per million tokens.
Cache Hit: Price per token for cached prompts (previously processed), typically offering a significant discount compared to regular input price, represented as USD per million tokens.
Cache Storage per Hour: Cost to maintain tokens in cache storage, charged per million tokens per hour. Currently only applicable to Google's Gemini models.
Performance Summary
Output Speed vs. Price
Output Speed: Output Tokens per Second; Price: USD per 1M Tokens
Most attractive quadrant
Output Speed: 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: Price per token, represented as USD per million Tokens. Price is a blend of Input & Output token prices (3:1 ratio).
Latency vs. Output Speed
Latency: Seconds to First Token Received; Output Speed: Output Tokens per Second
Most attractive quadrant
Size represents Price (USD per M Tokens)
Output Speed: 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).
Latency: Time to first token of tokens received, in seconds, after API request sent. For models which do not support streaming, this represents time to receive the completion.
Price: Price per token, represented as USD per million Tokens. Price is a blend of Input & Output token prices (3:1 ratio).
Median across providers: Figures represent median (P50) across all providers which support the model.
Speed
Measured by Output Speed (tokens per second)
Output Speed
Output Tokens per Second; Higher is better
Output Speed: 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).
Median across providers: Figures represent median (P50) across all providers which support the model.
Output Speed by Input Token Count (Context Length)
Output Tokens per Second; Higher is better
100 input tokens
1k input tokens
10k input tokens
100k input tokens
Output Speed: 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).
Input Tokens Length: Length of tokens provided in the request. See Prompt Options above to see benchmarks of different input prompt lengths across other charts.
Median across providers: Figures represent median (P50) across all providers which support the model.
Output Speed Variance
Output Tokens per Second; Results by percentile; Higher is better
Median, Other points represent 5th, 25th, 75th, 95th Percentiles respectively
Output Speed: 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).
Boxplot:Ā Shows variance of measurements
Output Speed, Over Time
Output Tokens per Second; Higher is better
Output Speed: 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).
Over time measurement: Median measurement per day, based on 8 measurements each day at different times. Labels represent start of week's measurements.
Latency
Measured by Time (seconds) to First Token
Latency
Seconds to First Token Received; Lower is better
Latency: Time to first token of tokens received, in seconds, after API request sent. For models which do not support streaming, this represents time to receive the completion.
Median across providers: Figures represent median (P50) across all providers which support the model.
Latency by Input Token Count (Context Length)
Seconds to First Token Received; Lower is better
100 input tokens
1k input tokens
10k input tokens
100k input tokens
Input Tokens Length: Length of tokens provided in the request. See Prompt Options above to see benchmarks of different input prompt lengths across other charts.
Latency: Time to first token of tokens received, in seconds, after API request sent. For models which do not support streaming, this represents time to receive the completion.
Median across providers: Figures represent median (P50) across all providers which support the model.
Latency Variance
Seconds to First Token Received; Results by percentile; Lower is better
Median, Other points represent 5th, 25th, 75th, 95th Percentiles respectively
Latency: Time to first token of tokens received, in seconds, after API request sent. For models which do not support streaming, this represents time to receive the completion.
Boxplot:Ā Shows variance of measurements
Latency, Over Time
Seconds to First Token Received; Lower is better
Latency: Time to first token of tokens received, in seconds, after API request sent. For models which do not support streaming, this represents time to receive the completion.
Over time measurement: Median measurement per day, based on 8 measurements each day at different times. Labels represent start of week's measurements.
Total Response Time
Time to receive 100 tokens output, calculated from latency and output speed metrics
Total Response Time
End-to-End Seconds to Output 100 Tokens; Lower is better
Total Response Time: Time to receive a 100 token response. Calculated based on Latency (time to receive first token) and Output Speed (output tokens per second).
Median across providers: Figures represent median (P50) across all providers which support the model.
Total Response Time by Input Token Count (Context Length)
End-to-End Seconds to Output 100 Tokens; Lower is better
100 input tokens
1k input tokens
10k input tokens
100k input tokens
Input Tokens Length: Length of tokens provided in the request. See Prompt Options above to see benchmarks of different input prompt lengths across other charts.
Total Response Time: Time to receive a 100 token response. Calculated based on Latency (time to receive first token) and Output Speed (output tokens per second).
Median across providers: Figures represent median (P50) across all providers which support the model.
Total Response Time Variance
Total Response Time: End-to-End Seconds to Output 100 Tokens; Results by percentile; Lower is better
Median, Other points represent 5th, 25th, 75th, 95th Percentiles respectively
Total Response Time: Time to receive a 100 token response. Calculated based on Latency (time to receive first token) and Output Speed (output tokens per second).
Boxplot:Ā Shows variance of measurements
Total Response Time, Over Time
End-to-End Seconds to Output 100 Tokens; Lower is better
Total Response Time: Time to receive a 100 token response. Calculated based on Latency (time to receive first token) and Output Speed (output tokens per second).
Over time measurement: Median measurement per day, based on 8 measurements each day at different times. Labels represent start of week's measurements.
Features | Price | Output tokens/s | Latency | ||||
|---|---|---|---|---|---|---|---|
Further Analysis | |||||||
 | GPT-4o (Aug '24) | 128k | 78 | $4.38 | 67.6 | 0.99 | |
| GPT-4o (May '24) | 128k | 78 | $7.50 | 118.1 | 0.74 | |
| GPT-4o mini | 128k | 73 | $0.26 | 171.0 | 0.77 | |
| GPT-4o (Nov '24) | 128k | 73 | $4.38 | 43.8 | 1.07 | |
| Llama 3.3 70B | 128k | 74 | $0.71 | 26.2 | 0.46 | |
| Llama 3.1 405B | 128k | 74 | $8.00 | 20.6 | 0.54 | |
| Llama 3.1 70B | 128k | 68 | $2.90 | 32.1 | 0.56 | |
| Llama 3.1 8B | 128k | 54 | $0.38 | 163.3 | 0.31 | |
| Mistral Large 2 (Jul '24) | 128k | 74 | $3.00 | 29.4 | 0.55 | |
| Mistral Large 2 (Nov '24) | 128k | 74 | $3.00 | 36.4 | 0.54 | |
| Command-R+ (Apr '24) | 128k | 44 | $6.00 | 48.8 | 0.59 | |
| Command-R (Mar '24) | 128k | 36 | $0.75 | 78.2 | 0.46 | |
| Phi-3 Medium 14B | 128k | $0.30 | 49.5 | 0.43 | ||
| Jamba 1.5 Large | 256k | 64 | $3.50 | 50.8 | 0.71 | |
| Jamba 1.5 Mini | 256k | $0.25 | 82.7 | 0.49 | ||
| GPT-4 Turbo | 128k | 75 | $15.00 | 47.9 | 1.48 | |
| Llama 3 8B | 8k | 45 | $0.38 | 73.7 | 0.39 | |
| Mistral Small (Feb '24) | 33k | 59 | $1.50 | 52.6 | 0.39 | |
| Mistral Large (Feb '24) | 33k | 55 | $6.00 | 35.4 | 0.52 | |
| Jamba Instruct | 256k | 28 | $0.55 | 75.8 | 0.53 |
Key definitions
Artificial Analysis Quality Index: Average result across our evaluations covering different dimensions of model intelligence. Currently includes MMLU, GPQA, Math & HumanEval. OpenAI o1 model figures are preliminary and are based on figures stated by OpenAI. See methodology for more details.
Context window: Maximum number of combined input & output tokens. Output tokens commonly have a significantly lower limit (varied by model).
Output Speed: 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).
Latency: Time to first token of tokens received, in seconds, after API request sent. For models which do not support streaming, this represents time to receive the completion.
Price: Price per token, represented as USD per million Tokens. Price is a blend of Input & Output token prices (3:1 ratio).
Output Price: Price per token generated by the model (received from the API), represented as USD per million Tokens.
Input Price: Price per token included in the request/message sent to the API, represented as USD per million Tokens.
Time period: Metrics are 'live' and are based on the past 14 days of measurements, measurements are taken 8 times a day for single requests and 2 times per day for parallel requests.