Skip to content

Changelog

The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.

Note

Downtimes and maintenance windows are communicated upon subscription. You can subscribe here.

[1.28.0] - 2026-08-24

Added

Added optional detect_best_audience dimensions to:
- POST /v1/videos/predicted-reactions/{file_id}
- POST /v1/images/predicted-reactions/{file_id}
Added the optional best_audience response field to:

[1.27.1] - 2026-07-21

Changed

Update documentation

[1.27.0] - 2026-07-21

Added

New POST endpoint to add/overwrite KPIs, Metadata and ExternalIds to the file_id.
- POST /v1/files/{file_id}/data
- Can be added multiple key values at the same time or remove all at the same endpoint (with set the kpi|metadata|externalIds to empty object {} in the request body)
- IF key is not provided in the request, no change will be made to that key in the file_id.

Changed

Endpoints using the new kpi data in input/output
Update swagger documentation to reflect the new KPI usage.

[1.26.3] - 2026-07-09

Changed

Map reactions-v3.x/v4.x/v5.0 jobs to recommendations-v3.0 in returned job info.
Update documentation to reflect the new model version mapping.

[1.26.2] - 2026-07-07

Changed

Update documentation

[1.26.1] - 2026-06-29

Changed

Update documentation

[1.26.0] - 2026-06-26

Changed

Add new swagger url: https://api.adverteyes.ai/swagger, api url: https://api.adverteyes.ai and monitoring url: https://adverteyes.betteruptime.com/ to the documentation
Update documentation

[1.25.1] - 2026-06-04

Fixed

Fixed POST /v1/files/register failing on presigned S3 URLs by detecting content type via GET when HEAD is not supported.

[1.25.0] - 2026-05-29

Added

OAuth 2.0 Bearer token authentication. All protected endpoints now accept Authorization: Bearer <access_token> in addition to x-api-key. Both methods are supported interchangeably but cannot be combined in the same request.
OAuth 2.0 / OpenID Connect discovery endpoints:
- GET /.well-known/oauth-authorization-server
- GET /.well-known/openid-configuration

[1.24.8] - 2026-05-14

Added

Added reactions-v4.1. Documentation of the model and parameters can be found in the Model documentation.
- Available at endpoints:

[1.24.5] - 2026-04-10

Fixed

Vulnerability fixes

[1.24.4] - 2026-04-09

Added

New property recommendation_details added to endpoints:
- GET /v1/insights/recommendations/{file_id}
- GET /v1/insights/recommendations/{file_id}/{campaign_id}

[1.24.3] - 2026-02-26

Fixed

Unstable model version of scenes intelligence v2 removed from documentation.

[1.24.3-rc1] - 2026-02-24

Fixed

Fixed an issue where the API in rare cases returned empty ScenesIntelligence and incorrectly marked the state as “finished” before processing completed.

[1.24.1] - 2026-02-12

Fixed

Fixed error at getting diagnostics scores with GET /v1/insights/scores/{file_id}/{campaign_id} endpoint

[1.24.0] - 2026-02-11

Changed

Updated API endpoint hostnames for the Adverteyes rebrand to adverteyes.ai.

[1.23.1] - 2026-02-06

Fixed

Fixed an issue where the API returned empty recommendations and incorrectly marked the state as “finished” before processing completed.

[1.23.0] - 2026-02-06

Added

Added new recommendations based on campaign objective recommendations-v3.0.
- Available at endpoints:

Changed

Previous alpha versions of recommendations models and endpoints are still available, but marked as deprecated.
- The full list of deprecated models is: [recommendations-v1.0, recommendations-v2.0]
- The full list of deprecated endpoints is:

[1.22.10-beta] - 2026-01-21

Changed

Swagger documentation minor fixes

[1.22.9-beta] - 2025-12-19

Changed

Diagnostics and scores are aligned with media types (Videos and Images)
Conversion - Swipes score has been renamed to Conversion - Interactions at diagnostics endpoints:
- GET /v1/insights/diagnostics/{file_id}
- GET /v1/insights/scores/{file_id}/{campaign_id}

[1.22.8-beta] - 2025-12-02

Changed

Added description and impact to diagnostics at insight endpoints:
- GET /v1/insights/diagnostics/{file_id}
- GET /v1/insights/scores/{file_id}/{campaign_id}

[1.22.7-beta] - 2025-11-25

Changed

Remove reactions-v4.0-alpha.1 and add reactions-v4.0. Documentation of the model and parameters can be found in the Model documentation.
- Available at endpoints:

[1.22.6-beta] - 2025-11-20

Changed

Swagger documentation minor fixes

[1.22.5-beta] - 2025-11-20

Added

Added reactions-v4.0-alpha.1. Documentation of the model and parameters can be found in the Model documentation.
- Available at endpoints:

[1.22.4-beta] - 2025-11-14

Fixed

Fixed issue with duplicated diagnostics list at GET /v1/insights/scores/{file_id}/{campaign_id} endpoint, when option all has been used for campaign_id query parameter

[1.22.3-beta] - 2025-11-05

Changed

Swagger documentation minor fixes

[1.22.2-beta] - 2025-09-10

Changed

Swagger documentation minor fixes at endpoint:
- POST /v1/images/image-intelligence/{file_id}

[1.22.1-beta] - 2025-09-10

Changed

Swagger documentation minor fixes at endpoint:
- POST /v1/videos/scenes-intelligence/{file_id}
- POST /v1/images/image-intelligence/{file_id}

[1.22.0-beta] - 2025-09-10

Added

added required request body with optional model_version parameter to the following endpoints:
- POST /v1/videos/recommendations/{file_id}
- POST /v1/images/recommendations/{file_id}

[1.21.1-beta] - 2025-08-27

Fixed

Swagger documentation minor fixes at endpoint:
- GET /v1/insights/diagnostics/{file_id}

[1.21.0-beta] - 2025-08-27

Changed

Modified the response of the following endpoints - version can be found at different place in the response:
- GET /v1/insights/diagnostics/{file_id}
- GET /v1/insights/scores/{file_id}/{campaign_id}

[1.20.0-beta] - 2025-07-30

Added

Added new Api endpoints with Image intelligence (similar to Scenes-intelligence for videos) and Media insights

[1.19.5-beta] - 2025-07-24

Changed

Small improvement to file access

Fix

Batch model parameters validation.
- If the model version was not provided, the default parameters were used for that record.
  - Symptom: empty params were returned with the result

[1.19.4-beta] - 2025-07-14

Changed

Update GET /v1/files/search endpoint:
- Add date range filtering for created_at and updated_at fields
- Add examples for the Swagger documentation

[1.19.3-beta] - 2025-06-25

Changed

API swagger documentation updated at endpoints:
- POST /v1/videos/predicted_reactions/{file_id}
- POST /v1/images/predicted_reactions/{file_id}

[1.19.2-beta] - 2025-06-18

Changed

Added branding area at ad level for GET /v1/videos/scenes-intelligence/{file_id} endpoint.

[1.19.1-beta] - 2025-06-04

Changed

API swagger documentation updated with proper error codes per endpoint
Error messages updated to be consistent per all API endpoints

[1.19.0-beta] - 2025-05-22

Changed

Changed the default model version to reactions-v3.4. Documentation of the models and parameters can be found in the documentation. Available at endpoints:
- POST /v1/videos/predicted_reactions/{file_id}
- POST /v1/batch/predicted-reactions
- Extra metrics available in the response at video and scene level:
  - negativity - maximum of negative reactions
  - reactions - maximum of all reactions
Search endpoint was reworked to support filtering for metadata and external_ids, also introduced paging support
- POST /v1/files/search

[1.18.14-beta] - 2025-05-14

Changed

Images and videos uploaded and processed via Adverteyes Dashboard, are available via API to account owner with api_key scoped for the target account
- Note: request_id in this case will be empty.

[1.18.13-beta] - 2025-05-08

Changed

API documentation minor updates

[1.18.12-beta] - 2025-05-08

Changed

Increase max file size from 100 MB to 4 GB

[1.18.11-beta] - 2025-05-06

Fix

Fix media registration from web url in Batch processing POST /v1/batch/predicted_reactions
Fix reactions-v3.2 media without audio processing

[1.18.10-beta] - 2025-04-30

Changed

API documentation minor updates

[1.18.9-beta] - 2025-04-30

Changed

Add ExternalIds and Metadata attribute possibility for endpoints with file registering capability

Added

GET endpoint to list media files by their external_ids
- POST /v1/files/register
- Parameters: External IDs for the files in the format: key1=value1&key2=value2&key3=value3
Added reactions-v3.3. Documentation of the model and parameters can be found in the Model documentation.
- Available at endpoints:
  - POST /v1/videos/predicted_reactions/{file_id}
  - POST /v1/batch/predicted-reactions
- Parameter changes compared to reactions-v3.0:
  - platform_format - replacing previous ad_format parameter. New values can be found in the Model documentation

[1.18.8-beta] - 2025-04-17

Changed

API documentation minor updates

[1.18.7-beta] - 2025-04-17

Changed

Countries list updated with disclaimer on reliability in the Countries documentation
API documentation minor updates

Added

Attention and Emotions section to explaining signals like Attention, Distraction, Happiness, Suprise, and Confusion in the Attention and Emotions documentation

[1.18.6-beta] - 2025-04-17

Changed

API documentation minor updates

[1.18.5-beta] - 2025-04-17

Changed

Improve heatmap generation performance and add {“heatmap_version”:”“v1”} optional body parameter to
- POST /v1/videos/heatmap/{file_id} endpoint
- POST /v1/images/heatmap/{file_id} endpoint

[1.18.1-beta] - 2025-03-17

Changed

Excluded unsupported countries from country parameters for reactions-v3 model version:
- Updated list of supported countries in the Countries documentation

[1.18.0-beta] - 2025-03-14

Changed

Changed the default model version to reactions-v3 for videos and images
Added support new values for country and age parameters for reactions-v3 model version:
- country - XX new parameter value for Global country code
- age - 18-60 new parameter value for GEN POP age range

[1.17.0-beta] - 2025-02-27

Added

Fixed/Improved consistency of processing state update for batch processing requests.

[1.16.0-beta] - 2025-02-18

Added

Add time_to_earliest_branding and emotional_tone_presence attribute at video level for GET /v1/videos/scenes-intelligence/{file_id} endpoint.

[1.14.1-beta] - 2025-02-12

Changed

The POST /v1/batch/predicted-reactions now accepts urls without extensions
The POST /v1/files/register no longer needs the file_type paramter

[1.14.0-beta] - 2025-02-06

Added

New POST endpoint to media files from external urls without uploading them
- POST /v1/files/register

[1.13.1-beta] - 2025-02-05

Added

fixed missing number_of_focal_points attribute at scenes level for GET /v1/videos/scenes-intelligence/{file_id} endpoint.

[1.13.0-beta] - 2025-02-04

Added

New GET endpoint to get quota status of processing requests.
- GET /v1/quotas

[1.11.0-beta] - 2025-01-09

Added

Added reactions-v3.2. Documentation of the model and parameters can be found in the Model documentation.
- Available at endpoints:
  - POST /v1/videos/predicted_reactions/{file_id}
  - POST /v1/batch/predicted-reactions
- Extra metrics available in the response at video and scene level:
  - negativity - maximum of negative reactions
  - reactions - maximum of all reactions
- Default weights of Attention Potential score are: Attention (0.65), Happiness (0.2), Confusion(-0.025), Contempt (-0.025) and Surprise (-0.10)
- Parameter changes compared to reactions-v3.0:
  - platform_format - replacing previous ad_format parameter. New values can be found in the Model documentation
  - country - XX new parameter value for Global GEN POP country code
  - age - 18-60 new parameter value for GEN POP age range

[1.9.0-beta] - 2024-11-06

Changed

Image predicted_reactions model versions
- All predicted_reactions model versions can be used for images and videos
  - So for example reactions-v3.0 can be used for images, too with the parameters
- Changed the default model version to reactions-v2.1-composite for images
  - For video it is the same as before (reactions-v2.1-composite)
Batch processing
- csv lowercase, uppercase, and mixed case headers and parameters are now supported
- csv url errors are now more informative and at least one error per line is returned in the response to help with debugging

[1.7.0-beta] - 2024-10-24

Added

added request_id to the response of POST /v1/{videos|images}/predicted-reactions/{file_id} and POST /v1/{videos|images}/recommendations/{file_id} endpoints
New GET endpoints to get history of processing requests.
- reaction prediction
  - GET /v1/videos/predicted-reactions/{file_id}/requests
  - GET /v1/videos/predicted-reactions/{file_id}/requests/{request_id}
- recommendations
  - GET /v1/videos/recommendations/{file_id}/requests
  - GET /v1/videos/recommendations/{file_id}/requests/{request_id}

[1.6.0-beta] - 2024-10-22

Added

New POST/GET endpoints to generating and retrieving scene information including timing, branding, description.
- GET /v1/videos/scenes-intelligence/{file_id}
- POST /v1/videos/scenes-intelligence/{file_id}

[1.5.7-beta] - 2024-10-18

Added

Added norms percentiles for the predicted reactions results via API. Available:
- with all models versions
- with all endpoints that return predicted reactions results
- mean and peak: 25th, 50th, 75th percentiles of each reaction
- at video and scene level

Changed

ad_format parameter for reactions-v3.0 changed at API and APP:
- from: stream_free_browsing, tv_focused_exposure , shorts_free_browsing, feed_free_browsing
- to: stream, tv, shorts, feed
- Adverteyes API documentation updated accordingly

[1.5.6-beta] - 2024-10-15

Fixed

Fixed account limits calculations for POST API endpoints

[1.5.4-beta] - 2024-10-10

Added

Added reactions-v3.0 parameters as input option. Documentation of the models and parameters can be found in the documentation. Available at endpoints:
- POST /v1/videos/predicted_reactions/{file_id}
- POST /v1/batch/predicted-reactions
Batch processing input became more flexible.
- Only url or uid is required for each record. The rest of the columns are optional.
- Order of the columns in the input file is not important anymore.
- See examples at the swagger documentation POST /v1/batch/predicted-reactions

Changed

Recommendation and heatmap can be processed for the media after the predicted reactions are already processed for the media.

[1.5.2-beta] - 2024-09-30

Fixed

Fixed processing error when calling POST /v1/images/recommendations/{file_id} without calling POST /v1/images/predict-reactions/{file_id} first

[1.5.1-beta] - 2024-09-19

Fixed

Fixed batch processing when using external urls

[1.5.0-beta] - 2024-09-16

Changed

Changed default model_version to reactions-v2.1-composite for the POST /v1/*/predicted_reactions methods
Increase max file size from 20MB to 100 MB

[1.4.1-beta] - 2024-09-04

Fixed

Improved the processing stability

[1.4.0-beta] - 2024-08-29

Added

Added new reaction predictions models reactions-v3.0 for video processing.
Batch processing input changes. uid,url,model_version,externalid
- model_version - optional parameter to specify the model version for the processing. If not provided, the default model version will be used for that record
- externalid - optional parameter to specify the external id for the processing. The external_id will be available in the response for the processed record

[1.3.2-beta] - 2024-08-12

Fixed

Fixed issue at POST /v1/*/predicted_reactions, when *_peak data for newest model is not recalculated if it was already processed by previous model versions

[1.3.1-beta] - 2024-08-08

Added

Added new reaction predictions models reactions-v1.1-composite and reactions-v2.1-composite. The full list of existing models is:
- reactions-v1.0 - Attention Potential is Attention Mean Index and only video channel is analyzed (fit for videos without audio and images)
- reactions-v1.1-composite - Same as reactions-v1.0, but Attention Potential is a weighted average of subscores with correlation
- reactions-v2.0 - Attention Potential is Attention Mean Index. Video and Audio channels are analyzed
- reactions-v2.1-composite - Same as reactions-v2.0, but Attention Potential is a weighted average of subscores with correlation
Improve parameter hierarchy on the prediction object:
- added new fields to the prediction object:
  - reaction_data - contains all reactions subscores. The object includes reaction and their values in a format {‘reaction_key’:value} for the following keys:
    - {reaction}_mean
    - {reaction}_mean_index
    - {reaction}_peak
    - {reaction}_peak_index where, {reaction} is one of the following reactions: attention, distraction, confusion, contempt, happiness, disgust, surprise
  - custom_scores - contains custom scores if target model outputs them
Introduced show_reaction_data query parameter for GET /images/predicted_reactions and GET /videos/predicted_reactions endpoints to show all reactions subscores

Deprecated

Marked as deprecated attention_index and attention_mean field at response from GET /images/predicted_reactions and GET /videos/predicted_reactions
- the values of attention_index and attention_mean are available at under new reaction_data object at same response

[1.2.0-beta] - 2024-07-31

Fixed

Fixed issue at POST /v1/images/recommendations, when images were treated as videos

[1.1.0-beta] - 2024-07-24

Added

Added initial version of Adverteyes API documentation. the documentation is available by the link
Added support of recommendations for images API references link

Fixed

Fixed wrong url from the response of GET /v1/file for images

Changed

Codec of generated heatmap video changed from VP9 to AVC1

Deprecated

Marked as deprecated general_assessment field at response from GET /images/predicted_reactions
- the field is available at GET /images/recommendations

[1.0.2-beta] - 2024-07-22

Add

Introduce support of multiple reactions prediction models with help of model_version parameter at the /videos/predicted_reactions endpoint
- Currently supported 2 models:
  - reactions-v1.0 - Predicts reactions based only on video/image input
  - reactions-v2.0 - Predicts reactions based on video and audio tracks, provides better results for videos with audio in comparison with reactions-v1.0

[0.5.4-alpha] - 2024-07-15

Add

Introduced Batch processing for reactions prediction. Up to 100 videos/images can be processed at once. API references link

[0.5.3-alpha] - 2024-07-11

Add

Adverteyes API extended with reaction processing for images API references link
Adverteyes API extended with heatmap processing for images API references link

Change

Special characters at file name will be replaced with ‘_’ at upload file request
Increased scalability and performance of Reaction Prediction processing
Audio processing service is improved