Amazon Titan: Quality, Generation Time & Price Analysis
Analysis of Amazon's models and comparison to other image models across key metrics including quality, generation time, and price.
Models compared include Playground v2.5, Stable Diffusion 3 Medium, Stable Diffusion XL 1.0, SDXL Lightning, Stable Diffusion 1.5, Stable Diffusion 2.1, Amazon Titan G1 (Standard), DALLE 2, DALLE 3 HD, DALLE 3, Stable Diffusion 1.6, Stable Diffusion 3 Large Turbo, Stable Diffusion 3 Large, Amazon Titan G1 v2 (Standard), Playground v3 (beta), Ideogram v2, FLUX.1 [pro], FLUX.1 [dev], Ideogram v2 Turbo, Ideogram v1, FLUX1.1 [pro], Recraft 20B, FLUX.1 [schnell], Stable Diffusion 3.5 Large, Stable Diffusion 3.5 Large Turbo, and Recraft V3
API providers compared include fal.ai, Replicate, Amazon Bedrock, OpenAI, Fireworks, Stability.ai, Together.ai, Playground AI, Ideogram, Black Forest Labs, and Recraft AI.
For further details, see our methodology page.
Highlights
Quality ELO
Quality ELO: ELO score of model in Image Arena, Higher is better
Generation Time
Generation time: Seconds to generate 1 image, Lower is better
Price
Price: USD per 1000 image generations, Lower is better
Summary Analysis
Quality vs. Price
Quality ELO: ELO score of model in Image Arena, Price: USD per 1000 image generations
Most attractive quadrant
Size represents Generation time: Seconds to generate 1 image
Quality ELO: Relative ELO score of the models as determined by >100,000 responses from users in Artificial Analysis' Image Arena. Some models may not be shown due to not yet having enough votes.
Price: Price per 1k images generated by the model. For detail on how we calculate price per image for providers price based on inference time or steps, see our methodology page.
Note on Midjourney: Midjourney does not have an API. We benchmark Midjourney as an API using ImagineAPI which serves as a tool to access the Midjourney Discord.
Quality vs. Generation Time
Quality ELO: ELO score of model in Image Arena, Generation time: Seconds to generate 1 image
Most attractive quadrant
Size represents Price: USD per 1000 image generations
Quality ELO: Relative ELO score of the models as determined by >100,000 responses from users in Artificial Analysis' Image Arena. Some models may not be shown due to not yet having enough votes.
Generation Time: Median time the provider takes to generate an image over the past 14 days of measurements. This includes downloading the image from the provider where a URL is provided rather than an image response.
Note on Midjourney: Midjourney does not have an API. We benchmark Midjourney as an API using ImagineAPI which serves as a tool to access the Midjourney Discord.
Generation Time vs. Price
Generation time: Seconds to generate 1 image, Price: USD per 1000 image generations
Most attractive quadrant
Generation Time: Median time the provider takes to generate an image over the past 14 days of measurements. This includes downloading the image from the provider where a URL is provided rather than an image response.
Price: Price per 1k images generated by the model. For detail on how we calculate price per image for providers price based on inference time or steps, see our methodology page.
Note on Midjourney: Midjourney does not have an API. We benchmark Midjourney as an API using ImagineAPI which serves as a tool to access the Midjourney Discord.
Quality
Quality ELO (Image Arena)
Quality ELO: ELO score of model in Image Arena, Higher is better
Quality ELO: Relative ELO score of the models as determined by >100,000 responses from users in Artificial Analysis' Image Arena. Some models may not be shown due to not yet having enough votes.
Arena Win Rate
Arena Win Rate: % Win rate in Image Arena, Higher is better
Win Rate: Proportion of time an image generated by the model was selected as preferred compared to the other image present in Artificial Analysis' Image Arena.
Participate in the Image Arena to contribute to the crowdsourced quality evaluations
Generation Time
Generation Time
Generation time: Seconds to generate 1 image, Lower is better
Generation Time: Median time the provider takes to generate an image over the past 14 days of measurements. This includes downloading the image from the provider where a URL is provided rather than an image response.
Note on Midjourney: Midjourney does not have an API. We benchmark Midjourney as an API using ImagineAPI which serves as a tool to access the Midjourney Discord.
Generation Time, Variance
Generation time: Seconds to generate 1 image, Results by percentile, Lower is better
Median, Other points represent 5th, 25th, 75th, 95th Percentiles respectively
Generation Time: Median time the provider takes to generate an image over the past 14 days of measurements. This includes downloading the image from the provider where a URL is provided rather than an image response.
Boxplot: Shows variance of measurements
Generation Time, Over Time
Generation time: Seconds to generate 1 image, Lower is better
Generation Time: Median time the provider takes to generate an image over the past 14 days of measurements. This includes downloading the image from the provider where a URL is provided rather than an image response.
Over time measurement: Median measurement per day, based on 4 measurements each day at different times. Labels represent start of week's measurements.
Price
Price
Price: USD per 1000 image generations, Lower is better
Price: Price per 1k images generated by the model. For detail on how we calculate price per image for providers price based on inference time or steps, see our methodology page.
Amazon Bedrock