PR #27355: fix: add gpt-5.1-mini and gpt-5.1-nano registry entries with supports_none_reasoning_effort

• Repository: BerriAI/litellm
• Author: danielnc
• State: open | merged: False
• Link: https://github.com/BerriAI/litellm/pull/27355

Description (problem / solution / changelog)

Relevant issues

Fixes #27351

What this fixes

Per OpenAI's parameter compatibility guide, the gpt-5.1 family — including gpt-5.1-mini and gpt-5.1-nano — supports temperature when reasoning_effort is set to 'none' (or omitted, since the API default is 'none' for the 5.1 family). LiteLLM's pre-flight check in OpenAIGPT5Config.map_openai_params correctly applies this for the bare gpt-5.1 slug because gpt-5.1 has supports_none_reasoning_effort: true in model_prices_and_context_window.json.

But gpt-5.1-mini and gpt-5.1-nano are absent from the registry entirely, so the gate falls back to strict rejection — litellm.UnsupportedParamsError even when reasoning_effort is null/omitted.

Reproduction (from #27351)

LiteLLM v1.83.10-stable, standard /v1/chat/completions (no Responses-API routing, no Azure):

Case A — openai/gpt-5.1 (control, in registry):

$ curl -sw "\n%{http_code}\n" -X POST $LITELLM_URL/v1/chat/completions \
    -H "Authorization: Bearer $VKEY" -H "Content-Type: application/json" \
    -d '{"model":"openai/gpt-5.1","messages":[{"role":"user","content":"reply ok"}],"temperature":0.7,"max_tokens":50}'
{"id":"chatcmpl-...","model":"openai/gpt-5.1","object":"chat.completion","choices":[...],...}
200

Case B — openai/gpt-5.1-mini (not in registry):

{"error":{"message":"litellm.UnsupportedParamsError: gpt-5 models (including gpt-5-codex) don't support temperature=0.7. Only temperature=1 is supported. For gpt-5.1, temperature is supported when reasoning_effort='none' (or not specified, as it defaults to 'none')...","type":"None","param":null,"code":"400"}}
400

Case C — openai/gpt-5.1-nano (not in registry): identical 400.

After this PR: B and C return 200, matching A.

Changes

Adds two new entries to model_prices_and_context_window.json — gpt-5.1-mini and gpt-5.1-nano — placed alphabetically right after gpt-5.1-codex-mini. The flag matrix mirrors the existing gpt-5.1 entry, in particular:

• supports_none_reasoning_effort: true — the load-bearing flag that gates the temperature pass-through
• supports_minimal_reasoning_effort: true
• supports_xhigh_reasoning_effort: false
• All supports_* capability flags (function_calling, parallel_function_calling, prompt_caching, reasoning, response_schema, system_messages, tool_choice, vision, web_search, pdf_input, native_streaming) align with the bare gpt-5.1 row.
• mode: "chat", supported_endpoints: ["/v1/chat/completions", "/v1/batch", "/v1/responses"].

Pricing rationale

OpenAI has not (to my knowledge) yet published distinct pricing for gpt-5.1-mini / gpt-5.1-nano. Rather than invent numbers, this PR anchors pricing to the gpt-5-mini / gpt-5-nano registry entries (the 5.0-generation siblings on the same scale tier from the same provider). When OpenAI publishes explicit 5.1-mini/-nano pricing, those rows can be tightened in a follow-up — and meanwhile cost tracking won't fall through to $0.

Specifically:

• gpt-5.1-mini pricing matches gpt-5-mini (input 2.5e-07, output 2e-06, with flex / priority tiers populated per the sibling).
• gpt-5.1-nano pricing matches gpt-5-nano (input 5e-08, output 4e-07, with flex tiers populated).
• Context window: max_input_tokens: 272000, max_output_tokens: 128000 — same as the rest of the gpt-5.1 family and the gpt-5-mini/-nano siblings.

Overlap with #22621

There is an open PR #22621 that adds pricing for gpt-5.1-mini (using the OpenAI mini pricing tier with 400K context / 128K max output) but does not address the missing reasoning-effort flags and so does not actually fix the UnsupportedParamsError users see today. This PR is filed independently with a tighter, JSON-only scope so maintainers can decide merge order; if #22621 lands first, the conflict will be a single key and easy to resolve (and our mini context window can adopt the 400K value if maintainers prefer it). If this PR lands first, #22621 reduces to a clean pricing tightening on top.

Sibling note (NOT changed in this PR)

gpt-5.1-codex is in the registry with supports_none_reasoning_effort: false. If Codex variants follow the same OpenAI contract as the rest of the 5.1 family, that's likely also wrong — but it's a behavior change to an existing row rather than an unambiguous gap, so it's deliberately out of scope here. Same for gpt-5.1-codex-mini and gpt-5.1-codex-max. Filing this PR for the absent rows only.

Tests

Extends tests/test_litellm/llms/openai/test_gpt5_transformation.py:

• test_gpt5_1_model_detection: adds two assertions that _supports_reasoning_effort_level("gpt-5.1-mini", "none") and _supports_reasoning_effort_level("gpt-5.1-nano", "none") both resolve to True via the registry.
• test_gpt5_1_mini_nano_temperature_with_reasoning_effort_none: new test mirroring the existing test_gpt5_1_temperature_with_reasoning_effort_none, asserting temperature pass-through for the two new slugs across [0.0, 0.7, 1.0, 1.5] with reasoning_effort='none'.

Pre-Submission checklist

• I have added testing in tests/test_litellm/ directory
• My PR's scope is as isolated as possible — JSON registry + a targeted test extension only
• No code logic changed; pure data/registry update plus matching test
• I have requested a Greptile review by commenting @greptileai and received a Confidence Score of at least 4/5 before requesting a maintainer review (will request right after this PR opens)

Type

Bug Fix

Changed files

• .circleci/config.yml (modified, +308/-47)
• .github/workflows/check-lazy-openapi-snapshot.yml (removed, +0/-75)
• .github/workflows/create-release.yml (modified, +4/-2)
• .github/workflows/test-unit-proxy-db.yml (modified, +1/-0)
• .gitignore (modified, +3/-2)
• Dockerfile (modified, +3/-3)
• Makefile (modified, +3/-0)
• README.md (modified, +1/-1)
• codecov.yaml (modified, +6/-1)
• cookbook/litellm-ollama-docker-image/requirements.txt (modified, +1/-1)
• deploy/charts/litellm-helm/templates/deployment.yaml (modified, +17/-0)
• deploy/charts/litellm-helm/tests/deployment_tests.yaml (modified, +5/-5)
• deploy/charts/litellm-helm/values.yaml (modified, +17/-6)
• docker/Dockerfile.alpine (modified, +1/-1)
• docker/Dockerfile.database (modified, +3/-3)
• docker/Dockerfile.dev (modified, +1/-1)
• docker/Dockerfile.health_check (modified, +1/-1)
• docker/Dockerfile.non_root (modified, +3/-5)
• docs/images/local-testing/hosted-vllm-custom-tool-local-test.png (added, +0/-0)
• docs/my-website/docs/providers/crusoe.md (added, +196/-0)
• enterprise/enterprise_hooks/banned_keywords.py (modified, +17/-9)
• enterprise/enterprise_hooks/google_text_moderation.py (modified, +4/-5)
• enterprise/enterprise_hooks/openai_moderation.py (modified, +3/-5)
• enterprise/litellm_enterprise/enterprise_callbacks/secret_detection.py (modified, +20/-44)
• enterprise/litellm_enterprise/proxy/auth/custom_sso_handler.py (modified, +32/-22)
• enterprise/litellm_enterprise/proxy/hooks/managed_files.py (modified, +193/-125)
• enterprise/litellm_enterprise/proxy/management_endpoints/project_endpoints.py (modified, +54/-29)
• enterprise/pyproject.toml (modified, +3/-3)
• litellm-js/proxy/package-lock.json (added, +2054/-0)
• litellm-js/proxy/package.json (modified, +3/-3)
• litellm-js/spend-logs/package-lock.json (modified, +4/-4)
• litellm-js/spend-logs/package.json (modified, +1/-1)
• litellm-proxy-extras/litellm_proxy_extras/migrations/20260501195714_managed_resource_team_owner/migration.sql (added, +20/-0)
• litellm-proxy-extras/litellm_proxy_extras/schema.prisma (modified, +11/-5)
• litellm-proxy-extras/pyproject.toml (modified, +3/-3)
• litellm/__init__.py (modified, +17/-1)
• litellm/_logging.py (modified, +3/-66)
• litellm/anthropic_beta_headers_config.json (modified, +2/-2)
• litellm/batches/batch_utils.py (modified, +75/-3)
• litellm/caching/caching.py (modified, +2/-1)
• litellm/caching/caching_handler.py (modified, +77/-22)
• litellm/caching/dual_cache.py (modified, +19/-0)
• litellm/caching/qdrant_semantic_cache.py (modified, +123/-71)
• litellm/caching/redis_cache.py (modified, +9/-1)
• litellm/caching/redis_semantic_cache.py (modified, +139/-33)
• litellm/constants.py (modified, +14/-3)
• litellm/cost_calculator.py (modified, +5/-1)
• litellm/files/main.py (modified, +24/-0)
• litellm/integrations/arize/_utils.py (modified, +38/-4)
• litellm/integrations/arize/arize_phoenix_client.py (modified, +14/-1)
• litellm/integrations/arize/arize_phoenix_prompt_manager.py (modified, +9/-2)
• litellm/integrations/azure_sentinel/azure_sentinel.py (modified, +111/-28)
• litellm/integrations/bitbucket/bitbucket_client.py (modified, +22/-3)
• litellm/integrations/bitbucket/bitbucket_prompt_manager.py (modified, +9/-2)
• litellm/integrations/custom_sso_handler.py (modified, +11/-0)
• litellm/integrations/gcs_bucket/gcs_bucket.py (modified, +9/-4)
• litellm/integrations/gcs_bucket/gcs_bucket_base.py (modified, +10/-3)
• litellm/integrations/gitlab/gitlab_prompt_manager.py (modified, +9/-2)
• litellm/integrations/langfuse/langfuse.py (modified, +35/-8)
• litellm/integrations/langfuse/langfuse_handler.py (modified, +3/-1)
• litellm/integrations/langfuse/langfuse_prompt_management.py (modified, +10/-9)
• litellm/integrations/langsmith.py (modified, +35/-10)
• litellm/integrations/opentelemetry.py (modified, +35/-18)
• litellm/integrations/prometheus.py (modified, +179/-55)
• litellm/integrations/prometheus_helpers/bounded_prometheus_series_tracker.py (added, +107/-0)
• litellm/integrations/prometheus_helpers/prometheus_api.py (modified, +23/-1)
• litellm/litellm_core_utils/cli_token_utils.py (modified, +17/-5)
• litellm/litellm_core_utils/cloud_storage_security.py (added, +175/-0)
• litellm/litellm_core_utils/custom_logger_registry.py (modified, +2/-0)
• litellm/litellm_core_utils/exception_mapping_utils.py (modified, +16/-5)
• litellm/litellm_core_utils/get_llm_provider_logic.py (modified, +12/-0)
• litellm/litellm_core_utils/initialize_dynamic_callback_params.py (modified, +10/-3)
• litellm/litellm_core_utils/litellm_logging.py (modified, +7/-2)
• litellm/litellm_core_utils/llm_request_utils.py (modified, +3/-3)
• litellm/litellm_core_utils/prompt_templates/factory.py (modified, +106/-11)
• litellm/litellm_core_utils/realtime_streaming.py (modified, +278/-30)
• litellm/litellm_core_utils/secret_redaction.py (added, +81/-0)
• litellm/litellm_core_utils/streaming_handler.py (modified, +1/-1)
• litellm/litellm_core_utils/url_utils.py (modified, +164/-4)
• litellm/llms/anthropic/batches/transformation.py (modified, +3/-1)
• litellm/llms/anthropic/chat/handler.py (modified, +48/-4)
• litellm/llms/anthropic/chat/transformation.py (modified, +539/-56)
• litellm/llms/anthropic/common_utils.py (modified, +12/-1)
• litellm/llms/anthropic/experimental_pass_through/adapters/handler.py (modified, +33/-4)
• litellm/llms/anthropic/experimental_pass_through/adapters/transformation.py (modified, +18/-5)
• litellm/llms/anthropic/experimental_pass_through/messages/transformation.py (modified, +62/-0)
• litellm/llms/anthropic/files/handler.py (modified, +5/-1)
• litellm/llms/anthropic/files/transformation.py (modified, +7/-3)
• litellm/llms/anthropic/skills/transformation.py (modified, +3/-1)
• litellm/llms/azure/azure.py (modified, +35/-6)
• litellm/llms/azure/cost_calculation.py (modified, +5/-1)
• litellm/llms/azure/image_edit/transformation.py (modified, +18/-0)
• litellm/llms/azure/image_generation/__init__.py (modified, +2/-0)
• litellm/llms/azure/image_generation/http_utils.py (added, +17/-0)
• litellm/llms/azure/realtime/handler.py (modified, +4/-2)
• litellm/llms/azure/responses/transformation.py (modified, +9/-2)
• litellm/llms/azure_ai/agents/handler.py (modified, +14/-4)
• litellm/llms/azure_ai/anthropic/transformation.py (modified, +21/-1)
• litellm/llms/azure_ai/cost_calculator.py (modified, +2/-0)
• litellm/llms/azure_ai/ocr/document_intelligence/transformation.py (modified, +22/-1)