Skip to content

Unable to perform listFoundationModels() from Bedrock #781

@Andrea-Scuderi

Description

@Andrea-Scuderi

Describe the bug

Unable to perform listFoundationModels() from Bedrock

To Reproduce
Steps to reproduce the behaviour:

import PackageDescription

let package = Package(
    name: "BedrockTest",
    platforms: [
        .macOS(.v15)
    ],
    dependencies: [
        .package(url: "https://github.com/soto-project/soto.git", from: "7.10.0")
    ],
    targets: [
        .executableTarget(
            name: "BedrockTest",
            dependencies: [
               .product(name: "SotoBedrock", package: "soto"),
            ]
        )
    ]
)
import AsyncHTTPClient
import Logging
import SotoBedrock
import SotoCore

@main
struct BedrockTest {
    static func main() async throws {
        let httpClient = HTTPClient(eventLoopGroupProvider: .singleton)

        let awsClient = AWSClient(credentialProvider: .configFile(profile: "bedrock-test"), httpClient: httpClient)
        let bedrock = SotoBedrock.Bedrock(client: awsClient, region: .useast1)
        do {
            let result = try await bedrock.listFoundationModels()
            print("Foundation Models: \(result.modelSummaries)")
        } catch {
            print("Error listing foundation models: \(error)")
        }
        try? await awsClient.shutdown()
        try? await httpClient.shutdown()
    }
}

Error:

Error listing foundation models: dataCorrupted(Swift.DecodingError.Context(codingPath: [CodingKeys(stringValue: "modelSummaries", intValue: nil), _CodingKey(stringValue: "Index 0", intValue: 0), CodingKeys(stringValue: "inferenceTypesSupported", intValue: nil), _CodingKey(stringValue: "Index 0", intValue: 0)], debugDescription: "Cannot initialize InferenceType from invalid String value INFERENCE_PROFILE", underlyingError: nil))

From aws cli the output is:

aws bedrock list-foundation-models --profile bedrock-test --region us-east-1
{
    "modelSummaries": [
        {
            "modelArn": "arn:aws:bedrock:us-east-1::foundation-model/stability.stable-fast-upscale-v1:0",
            "modelId": "stability.stable-fast-upscale-v1:0",
            "modelName": "Stable Image Fast Upscale",
            "providerName": "Stability AI",
            "inputModalities": [
                "TEXT",
                "IMAGE"
            ],
            "outputModalities": [
                "IMAGE"
            ],
            "responseStreamingSupported": false,
            "customizationsSupported": [],
            "inferenceTypesSupported": [
                "INFERENCE_PROFILE"
            ],
            "modelLifecycle": {
                "status": "ACTIVE"
            }
        },
...
}

Expected behavior
A clear and concise description of what you expected to happen.

according to this https://docs.aws.amazon.com/bedrock/latest/APIReference/API_ListFoundationModels.html
the content of inferenceTypesSupported array should be of type string.

in soto, the shape is defined as an enum, but INFERENCE_PROFILE is not present in the case list:

 public struct FoundationModelDetails: AWSDecodableShape {
        /// The customization that the model supports.
        public let customizationsSupported: [ModelCustomization]?
        /// The inference types that the model supports.
        public let inferenceTypesSupported: [InferenceType]?

        ...

public enum InferenceType: String, CustomStringConvertible, Codable, Sendable, CodingKeyRepresentable {
        case onDemand = "ON_DEMAND"
        case provisioned = "PROVISIONED"
        public var description: String { return self.rawValue }
 }

Setup (please complete the following information):

  • OS: macOS Tahoe 26.0.1
  • Version of soto: 7.10
  • Authentication mechanism [hard-coded credentials, IAM Instance Profile on EC2, etc] configuration file

Additional context

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions