Repository Issue Activity (beta)

#798 Dataset exports leave streamed HTTP responses open

#796 Batched embeddings drop types absent from the first response

#777 Client.embed(texts=[]) raises IndexError instead of handling empty input

#14 automate tests in python sdk

#651 Properly return usage information for BedrockClientV2 for the Reranker model?

#762 Bug: uncaught type error in _get_tokenizer_config_size when server omits content length

#774 Bug: construct_type crashes with ValueError on bare unparameterized dict

#770 BUG: TypeError in merge_embed_responses when later batches have None embedding fields

#772 BUG: AttributeError in sum_fields_if_not_none when billed_units contains None entries

#735 Feature request: Oracle Cloud Infrastructure (OCI) Generative AI client support

#760 fix: OCI vision requests fail — image_url not converted to camelCase

#759 fix: OCI tool use returns 400 — tool type field not uppercased

#756 fix: OCI streaming hangs indefinitely — OCI does not send SSE [DONE] marker

#741 Upgrade tokenizers dependency to get rid of CVE-2026-4539

#734 Feature request: HTTP connection pooling for improved performance

#719 embedding_type may be mandatory in some cases but is marked as optional.

#534 Allow users to configure `embed_batch_size` or `ThreadPoolExecutor` size when calling `Client.embed`

#733 Feature request: Memory-efficient streaming API for embed()

#493 Offline tokenization produces empty token_strings

#730 chat_stream() logs spurious warning "Failed to parse SSE data field as JSON" for [DONE] stream terminator

#721 BedrockClientV2 SigV4 Signing Failure (Host Mismatch) and cohere_aws Dependency Issues

#706 cohere, Pydantic v1 3.14 error

#568 Include license details in package metadata

#592 Cannot stream response from cohere

#591 Streaming comes with extra data.