Model components / Fine-tuning code

Amazon Bedrock Custom Models / Fine-Tuning

Managed fine-tuning for Anthropic Claude 3 Haiku, Meta Llama, Cohere Command, Amazon Titan, and Amazon Nova models inside Bedrock, the only fully managed service offering fine-tuning of Claude weights. Picked when teams are already on AWS and need governance, VPC isolation, and Provisioned Throughput serving for the resulting custom model. GA since mid-2024; usage requires Provisioned Throughput commitment for inference.

Amazon Bedrock model customization (docs live June 2026). Methods: supervised fine-tuning, reinforcement fine-tuning (Lambda-defined reward functions), and distillation. Customizes Bedrock foundation models. Proprietary managed AWS service.

Openness

1 high confidence

1.0

source: closed
training-code: closed
runs-on: AWS-Bedrock-only
license: Proprietary(managed AWS service, token+storage billed)

Proprietary managed customization service inside AWS Bedrock, no source, runs only on AWS against Bedrock-hosted foundation models.

https://docs.aws.amazon.com/bedrock/latest/userguide/custom-models.html recorded 2026-06-04
supervised fine-tuning, reinforcement fine-tuning, distillation; managed AWS service; token/storage pricing

Adoption

not assessed

No disclosed usage figure for the Bedrock customization feature located this run (jobs run / customers using fine-tuning). Bedrock overall is widely used, but no honest signal specific to the custom-models/fine-tuning capability was found; declining to assign a level.

https://docs.aws.amazon.com/bedrock/latest/userguide/custom-models.html recorded 2026-06-04
managed customization feature; no usage figures disclosed

Capability

3 medium confidence

3.0

Solid managed method set (SFT + RFT + distillation), but a black-box service with no exposed parallelism/precision/scale controls and no benchmark basis. Scored 3: capable managed FT across a model catalog, below the RFT-on-frontier-reasoning depth of the OpenAI/Vertex offerings and far below the OSS scale-frontier libs.

https://docs.aws.amazon.com/bedrock/latest/userguide/custom-models.html recorded 2026-06-04
SFT, reinforcement fine-tuning with Lambda reward funcs, distillation

Unchanged since 2026-06-24 (last edited, not re-checked)