PR #25395: fix(embedding): omit null encoding_format for openai requests

• Repository: BerriAI/litellm
• Author: meutsabdahal
• State: closed | merged: True
• Link: https://github.com/BerriAI/litellm/pull/25395

Description (problem / solution / changelog)

Relevant issues

Fixes #25388

Pre-Submission checklist

Please complete all items before asking a LiteLLM maintainer to review your PR

• I have Added testing in the tests/test_litellm/ directory, Adding at least 1 test is a hard requirement - see details
• My PR passes all unit tests on make test-unit
• My PR's scope is as isolated as possible, it only solves 1 specific problem
• I have requested a Greptile review by commenting @greptileai and received a Confidence Score of at least 4/5 before requesting a maintainer review

Delays in PR merge?

If you're seeing a delay in your PR being merged, ping the LiteLLM Team on Slack (#pr-review).

CI (LiteLLM team)

CI status guideline:

• 50-55 passing tests: main is stable with minor issues.
• 45-49 passing tests: acceptable but needs attention
• <= 40 passing tests: unstable; be careful with your merges and assess the risk.

• Branch creation CI run
  Link:

• CI run for the last commit
  Link:

• Merge / cherry-pick CI run
  Links:

Type

🐛 Bug Fix ✅ Test

Changes

• Stop sending encoding_format: null in embedding requests when it is not set.
• Preserve explicit encoding_format values (for example, "float").
• Add regression tests for both behaviors (unset omitted, explicit preserved).

Changed files

• litellm/main.py (modified, +0/-3)
• tests/test_litellm/llms/openai/embeddings/test_openai_embeddings_encoding_format.py (added, +34/-0)

PR #25428: fix(embeddings): only set encoding_format=None for canonical OpenAI provider

• Repository: BerriAI/litellm
• Author: avarga1
• State: open | merged: False
• Link: https://github.com/BerriAI/litellm/pull/25428

Description (problem / solution / changelog)

Changes

Fix — `encoding_format=null` sent to third-party providers (closes #25388)

`litellm/main.py` unconditionally set `optional_params["encoding_format"] = None` when no `encoding_format` was provided, even for providers like `together_ai` and `nvidia_nim` that share the same embedding code path. These providers reject the `null` value with a 400 error (e.g. Gitee AI, SiliconFlow).

The `None` sentinel exists to prevent the OpenAI Python SDK from injecting its own default of `"float"` — but this is only relevant for the canonical OpenAI provider. Scoping it to `custom_llm_provider == "openai"` (or `None` for the `open_ai_embedding_models` fallback) leaves third-party providers unaffected.

# before — None sent to all providers in this branch
if encoding_format is not None:
    optional_params["encoding_format"] = encoding_format
else:
    # Omiting causes openai sdk to add default value of "float"
    optional_params["encoding_format"] = None

# after — None only set for canonical OpenAI
if encoding_format is not None:
    optional_params["encoding_format"] = encoding_format
elif custom_llm_provider == "openai" or custom_llm_provider is None:
    # Omitting causes openai sdk to add default value of "float".
    # Only suppress for the canonical OpenAI provider — third-party
    # providers (together_ai, nvidia_nim, etc.) reject encoding_format=null.
    optional_params["encoding_format"] = None

Test plan

• together_ai embedding call with no encoding_format — encoding_format key absent from params sent to provider
• nvidia_nim embedding call with no encoding_format — encoding_format key absent from params
• openai embedding call with no encoding_format — encoding_format=None still set (preserves existing SDK behaviour)
• Explicit encoding_format="float" always forwarded regardless of provider

Changed files

• litellm/main.py (modified, +4/-2)
• tests/test_litellm/test_utils.py (modified, +59/-16)