[Bug]:

[bwthomas]

Bug Description

Official AWS documentation indicates that inference is supported with haiku 4.5, and yet attempting to specify this model with the correct identifier, us.anthropic.claude-haiku-4-5-20251001-v1:0, gives the error Model does not support inference profiles but has an inference profile prefix.

Version

0.12.6

Steps to Reproduce

Specify an LLM call using the identifier us.anthropic.claude-haiku-4-5-20251001-v1:0

Relevant Logs/Tracbacks

[dosubot]

Hi @bwthomas! I'm Dosu and I’m helping the llama_index team.

The error happens because LlamaIndex 0.12.6 doesn't support inference profile prefixes (like us.) for Anthropic Claude Haiku 4.5 models. Support for these region/inference profile prefixes and the correct parsing of such model identifiers was added in later versions of both the core package and the relevant integration packages. To resolve this, upgrade both llama-index and llama-index-llms-anthropic to the latest versions—this will enable support for the model identifier format you're using and fix the inference profile error. There isn't a workaround for this in 0.12.6, as the necessary implementation simply isn't present in that version. For more details, see this related issue and recent PRs that added region/inference profile support.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}

   

[logan-markewich]

Amazon is weird with the model names. This error is coming directly from the bedrock api. Ive found that I need to sometimes include the arn in the model name

[logan-markewich]

For example:

model="arn:aws:bedrock:us-east-2:<arn_profile_num>:inference-profile/us.anthropic.claude-3-7-sonnet-20250219-v1:0"

[bwthomas]

If I understood the advice, updating to the most recent version for the packages named (llama-index & llama-index-llms-anthropic) should fix the issue. I've ensured that each package is on the latest version, and I still get the same issue. Versions are as follows:

> pip show llama-index | grep "^Version"
Version: 0.14.13
> pip show llama-index-llms-anthropic | grep "^Version"
Version: 0.10.7
> pip show llama-index-llms-bedrock-converse | grep "^Version"
Version: 0.12.6

Thank you for the suggestion, @logan-markewich. Unfortunately this still yields an error, albeit slightly different, which I've included below. Interestingly enough, haiku 4.5 isn't included in the list of known models, so ... I dunno 🤷

Note: AWS ID is changed to protect, well, me I guess? 🤣

[ERROR] Unknown model: arn:aws:bedrock:us-east-1:123456789012:inference-profile/us.anthropic.claude-
haiku-4-5-20251001-v1:0. Please provide a valid Bedrock model name. Known models are: amazon.nova-pro-v1:0, 
amazon.nova-lite-v1:0, amazon.nova-micro-v1:0, amazon.titan-text-express-v1, amazon.titan-text-lite-v1, amazon.titan-text-
premier-v1:0, anthropic.claude-instant-v1, anthropic.claude-v1, anthropic.claude-v2, anthropic.claude-v2:1, 
anthropic.claude-3-sonnet-20240229-v1:0, anthropic.claude-3-haiku-20240307-v1:0, anthropic.claude-3-opus-20240229-
v1:0, anthropic.claude-3-5-sonnet-20240620-v1:0, anthropic.claude-3-5-sonnet-20241022-v2:0, anthropic.claude-3-5-
haiku-20241022-v1:0, anthropic.claude-3-7-sonnet-20250219-v1:0, ai21.j2-mid-v1, ai21.j2-ultra-v1, cohere.command-text-
v14, cohere.command-light-text-v14, cohere.command-r-v1:0, cohere.command-r-plus-v1:0, meta.llama2-13b-chat-v1, 
meta.llama2-70b-chat-v1, meta.llama3-8b-instruct-v1:0, meta.llama3-70b-instruct-v1:0, meta.llama3-1-8b-instruct-v1:0, 
meta.llama3-1-70b-instruct-v1:0, meta.llama3-2-1b-instruct-v1:0, meta.llama3-2-3b-instruct-v1:0, meta.llama3-2-11b-
instruct-v1:0, meta.llama3-2-90b-instruct-v1:0, meta.llama3-3-70b-instruct-v1:0, mistral.mistral-7b-instruct-v0:2, 
mistral.mixtral-8x7b-instruct-v0:1, mistral.mistral-large-2402-v1:0, mistral.mistral-small-2402-v1:0, mistral.mistral-
large-2407-v1:0, ai21.jamba-1-5-mini-v1:0, ai21.jamba-1-5-large-v1:0, deepseek.r1-v1:0

[mattref]

The Amazon Bedrock Model ID for Haiku 4.5 is anthropic.claude-haiku-4-5-20251001-v1:0. If you're using the us. region prefix for Cross-region Inference, then just specify us.anthropic.claude-haiku-4-5-20251001-v1:0 as the model when using BedrockConverse. You don't need to pass an ARN to model. I suspect you were just on an outdated version of llama-index-llms-bedrock-converse