Menu

logo
Artificial Analysis
HOME
logo

Amazon Bedrock: Models Quality, Performance & Price

Analysis of Amazon Bedrock'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 Amazon Bedrock for your use-case. For more details including relating to our methodology, see our FAQs. Models analyzed: Llama 3.3 70B, Llama 3.1 405B Standard, Llama 3.1 405B Latency Optimized, Llama 3.1 70B Standard, Llama 3.1 70B Latency Optimized, Llama 3.2 90B (Vision), Llama 3.2 11B (Vision), Llama 3.1 8B, Llama 3.2 3B, Llama 3.2 1B, Claude 3.5 Sonnet (Oct), Claude 3.5 Sonnet (June), Claude 3 Opus, Claude 3.5 Haiku Standard, Claude 3.5 Haiku Latency Optimized, Claude 3 Haiku, Mistral Large 2 (Jul '24), Mixtral 8x7B, Command-R+, Command-R+ (Apr '24), Command-R (Mar '24), Command-R, Nova Pro, Nova Lite, Nova Micro, Llama 3 70B, Llama 3 8B, Claude 3 Sonnet, Claude 2.1, Mistral Large (Feb '24), and Mistral 7B.
Link:

Amazon Model Comparison Summary

Quality:Nova Pro logo Nova ProĀ andĀ Nova Lite logo Nova LiteĀ are the highest quality models offered by Amazon, followed by Llama 3.2 90B (Vision) logo Llama 3.2 90B (Vision), Nova Micro logo Nova Micro & Llama 3.2 11B (Vision) logo Llama 3.2 11B (Vision).Output Speed (tokens/s):Llama 3.2 1B logo Llama 3.2 1B (314 t/s)Ā andĀ Nova Micro logo Nova Micro (195 t/s)Ā are the fastest models offered by Amazon, followed by Nova Lite logo Nova Lite, Llama 3.2 3B logo Llama 3.2 3B & Llama 3.1 70B Latency Optimized logo Llama 3.1 70B Latency Optimized.Latency (seconds):Nova Micro logo Nova Micro (0.32s)Ā and Ā Llama 3 8B logo Llama 3 8B (0.33s)Ā are the lowest latency models offered by Amazon, followed by Mistral 7B logo Mistral 7B, Nova Lite logo Nova Lite & Mixtral 8x7B logo Mixtral 8x7B.Blended Price ($/M tokens):Nova Micro logo Nova Micro ($0.06)Ā andĀ Llama 3.2 1B logo Llama 3.2 1B ($0.10)Ā are the cheapest models offered by Amazon, followed by Nova Lite logo Nova Lite, Llama 3.2 3B logo Llama 3.2 3B & Llama 3.2 11B (Vision) logo Llama 3.2 11B (Vision).Context Window Size:Nova Pro logo Nova Pro (300k)Ā andĀ Nova Lite logo Nova Lite (300k)Ā are the largest context window models offered by Amazon, followed by Claude 3.5 Sonnet (Oct) logo Claude 3.5 Sonnet (Oct), Claude 3.5 Sonnet (June) logo Claude 3.5 Sonnet (June) & Claude 3 Opus logo Claude 3 Opus.

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: 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

ModelsFunction callingJSON Mode
Claude 3.5 Sonnet (Oct), Amazon logoClaude 3.5 Sonnet (Oct), Amazon
Claude 3.5 Sonnet (June), Amazon logoClaude 3.5 Sonnet (June), Amazon
Nova Pro, Amazon logoNova Pro, Amazon
Llama 3.3 70B, Amazon logoLlama 3.3 70B, Amazon
Mistral Large 2 (Jul '24), Amazon logoMistral Large 2 (Jul '24), Amazon
Llama 3.1 405B Standard, Amazon logoLlama 3.1 405B Standard, Amazon
Llama 3.1 405B Latency Optimized, Amazon logoLlama 3.1 405B Latency Optimized, Amazon
Nova Lite, Amazon logoNova Lite, Amazon
Claude 3 Opus, Amazon logoClaude 3 Opus, Amazon
Claude 3.5 Haiku Standard, Amazon logoClaude 3.5 Haiku Standard, Amazon
Claude 3.5 Haiku Latency Optimized, Amazon logoClaude 3.5 Haiku Latency Optimized, Amazon
Llama 3.1 70B Standard, Amazon logoLlama 3.1 70B Standard, Amazon
Llama 3.1 70B Latency Optimized, Amazon logoLlama 3.1 70B Latency Optimized, Amazon
Nova Micro, Amazon logoNova Micro, Amazon
Llama 3.1 8B, Amazon logoLlama 3.1 8B, Amazon
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).

Quality vs. Price

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)
ModelsCache Pricing Notes
Amazon Bedrock logoAmazon Bedrock
  • Amazon supports caching for Nova models and Anthropic's Claude models.
  • Pricing and usage differs between model families.
Claude 3.5 Sonnet (Oct) logoClaude 3.5 Sonnet (Oct)
  • Only supported on US West (Oregon) region
Claude 3.5 Sonnet (June) logoClaude 3.5 Sonnet (June)
  • Only supported on US West (Oregon) region
Nova Pro logoNova Pro
  • Only supported on US East (N. Virginia) region
Nova Lite logoNova Lite
  • Only supported on US East (N. Virginia) region
Claude 3.5 Haiku Standard logoClaude 3.5 Haiku Standard
  • Only supported on US West (Oregon) region
Claude 3.5 Haiku Latency Optimized logoClaude 3.5 Haiku Latency Optimized
  • Only supported on US West (Oregon) region
Nova Micro logoNova Micro
  • Only supported on US East (N. Virginia) region
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: 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

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
Picture of the author

Output Speed, Over Time

Output Tokens per Second; Higher is better
Smaller, emerging providers offer high output speed, though precise speeds delivered vary day-to-day.
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
Picture of the author

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
Picture of the author

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
Amazon Bedrock logo
Meta logo
Llama 3.3 70B
128k
74
$0.71
31.1
0.95
Amazon Bedrock Standard logo
Meta logo
Llama 3.1 405B Standard
128k
74
$2.40
30.5
1.96
Amazon Bedrock Latency Optimized logo
Meta logo
Llama 3.1 405B Latency Optimized
128k
74
$3.00
65.0
0.81
Amazon Bedrock Standard logo
Meta logo
Llama 3.1 70B Standard
128k
68
$0.72
31.5
0.71
Amazon Bedrock Latency Optimized logo
Meta logo
Llama 3.1 70B Latency Optimized
128k
68
$0.90
133.2
0.42
Amazon Bedrock logo
Meta logo
Llama 3.2 90B (Vision)
128k
67
$0.72
35.7
0.51
Amazon Bedrock logo
Meta logo
Llama 3.2 11B (Vision)
128k
53
$0.16
132.7
0.36
Amazon Bedrock logo
Meta logo
Llama 3.1 8B
128k
54
$0.22
90.6
0.39
Amazon Bedrock logo
Meta logo
Llama 3.2 3B
128k
49
$0.15
143.6
0.41
Amazon Bedrock logo
Meta logo
Llama 3.2 1B
128k
26
$0.10
313.6
0.35
Amazon Bedrock logo
Anthropic logo
Claude 3.5 Sonnet (Oct)
200k
80
$6.00
42.0
1.12
Amazon Bedrock logo
Anthropic logo
Claude 3.5 Sonnet (June)
200k
76
$6.00
45.5
1.03
Amazon Bedrock logo
Anthropic logo
Claude 3 Opus
200k
70
$30.00
23.6
1.51
Amazon Bedrock Standard logo
Anthropic logo
Claude 3.5 Haiku Standard
200k
68
$1.60
56.3
0.75
Amazon Bedrock Latency Optimized logo
Anthropic logo
Claude 3.5 Haiku Latency Optimized
200k
68
$2.00
101.0
0.59
Amazon Bedrock logo
Anthropic logo
Claude 3 Haiku
200k
55
$0.50
110.5
0.79
Amazon Bedrock logo
Mistral logo
Mistral Large 2 (Jul '24)
128k
74
$3.00
34.4
0.47
Amazon Bedrock logo
Mistral logo
Mixtral 8x7B
33k
41
$0.51
71.0
0.35
Amazon Bedrock logo
Cohere logo
Command-R+
128k
55
$6.00
47.6
0.52
Amazon Bedrock logo
Cohere logo
Command-R+ (Apr '24)
128k
45
$6.00
47.3
0.51
Amazon Bedrock logo
Cohere logo
Command-R (Mar '24)
128k
36
$0.75
108.7
0.36
Amazon Bedrock logo
Cohere logo
Command-R
128k
$0.75
108.4
0.36
Amazon Bedrock logo
Amazon logo
Nova Pro
300k
75
$1.40
89.7
0.37
Amazon Bedrock logo
Amazon logo
Nova Lite
300k
70
$0.10
145.7
0.33
Amazon Bedrock logo
Amazon logo
Nova Micro
130k
66
$0.06
194.8
0.32
Amazon Bedrock logo
Meta logo
Llama 3 70B
8k
47
$2.86
47.1
0.45
Amazon Bedrock logo
Meta logo
Llama 3 8B
8k
45
$0.38
102.4
0.33
Amazon Bedrock logo
Anthropic logo
Claude 3 Sonnet
200k
57
$6.00
52.6
0.78
Amazon Bedrock logo
Anthropic logo
Claude 2.1
200k
$12.00
29.1
1.80
Amazon Bedrock logo
Mistral logo
Mistral Large (Feb '24)
33k
56
$6.00
43.6
0.41
Amazon Bedrock logo
Mistral logo
Mistral 7B
8k
28
$0.16
92.6
0.33

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.