NewYour coding agent can read the release notes before it upgrades.Set up the MCP server →
PyPI
PyPI, #964 most downloaded on PyPI
MLflow Tracing SDK is an open-source, lightweight Python package that only includes the minimum set of dependencies and functionality to instrument your code/models/agents with MLflow Tracing.
Latest stable
3.18.0
Last release 18 days ago
03 Nov 2026
Ships fairly regularly
a new release about every 2 weeks
Nearly every release is documented
notes for 31 of 31 stable releases
2 versions withdrawn
withdrawn after publishing
2 years old
48 releases · first in 2025
Drag across the ruler to compare, or tap a release to open it.
[Tracing / Tracking] Fix FileStore trace path traversal (CVE-2026-53995) (#26842, @SaiTeja-Erukude)
MLflow 3.18.0 helps you reuse agent expertise, find regressions before shipping, and monitor quality in production.
search_logged_models metrics by the highest step, then latest timestamp, on SQL stores; apply datasets to both filtering and sorting (#27215, @agtzrubio)registered_model_meta to models:/ downloads; custom readers should use the URI or Model Registry (#27173, @sunilmehta-si)Database upgrade: Back up the SQL backend and plan a maintenance window. Stop MLflow servers and other writers, run mlflow db upgrade <database-url> with MLflow 3.18.0, and restart on the new version. The MCP version migration checks MySQL and SQL Server references for exact-case matches and stops before its own schema changes if correction is required. On case-insensitive databases, registry and MCP permissions now use stored resource names. Re-save differently cased grants under the stored spelling. Follow the database upgrade guidance.
MLFLOW_ALLOW_MODEL_CODE_EXECUTION=false; deprecate MLFLOW_ALLOW_PICKLE_DESERIALIZATION while preserving its pickle-only behavior (#27000, @joshuawong-db)quality_criteria parameter to mlflow.genai.evaluate (#26471, @joshuawong-db)session_level (#26443, @alkispoly-db)session_id to be a list in update_current_trace and tracing.context to identify hierarchical sessions (#26413, @alkispoly-db)TraceAudio to custom trace views (#26452, @chris-chieu)Bug fixes:
users/list to the requested workspace (#27270, @a-gajam)message_delta (#26908, @pcbeingused333)CapacityLimiter lazily to support anyio < 4.2 (#27405, @mlflow-app)wasbs:// URIs (#27396, @kriscon-db)AGENT spans at store time and record Strands cache tokens (#27371, @mlflow-app)MLFLOW_MODEL_CATALOG_URI before litellm in trace cost calculation (#27369, @mlflow-app)DeepEvalScorer returning another row's results under concurrent scoring (#27368, @mlflow-app)data: [DONE] (#27340, @mlflow-app)compute_pandas_digest failure on empty Arrow-backed string columns (#27359, @mlflow-app)_find_validator missing GetWorkspace validator for concrete workspace paths (#27355, @mlflow-app)bedrock:/ judges without a trace raising NotImplementedError in GatewayAdapter (#27322, @mlflow-app)POST writes that the server may have already processed (#27329, @kriscon-db)log_spans usage recompute (#27255, @rhamzatov)mlflow.groq.autolog (#27294, @mlflow-app)spark_udf environment archives (#26668, @B-Step62)SqlAlchemyStore with FileStore (#27311, @kriscon-db)FileStore.start_trace assessment handling with SqlAlchemyStore (#27309, @kriscon-db)databricks- model names from the Databricks catalog even when a provider is set (#27282, @kriscon-db)error event for Anthropic passthrough streaming failures (#25978, @rhamzatov)log_table (#26936, @sushu0)gateway:/ judge guardrails in the server process (#27147, @joshuawong-db)search_traces promising a page behind a result set that ends evenly (#27260, @tanghaoji). path segments when authorizing artifact proxy requests (#27229, @joshuawong-db)MLFLOW_ALLOW_MODEL_CODE_EXECUTION when loading LlamaIndex settings (#27231, @joshuawong-db)CODEX_API_KEY (#27074, @kriscon-db)BuiltInScorer subclasses when deserializing built-in scorers (#27035, @joshuawong-db)EvaluationDataset.delete_records() (#27247, @mlflow-app)register_model when Databricks Model Serving can't serve the model (#27161, @erarpitagarwal)roles/list and create_role workspaces; use sender_is_admin in GraphQL (#27171, @a-gajam)CASCADE workspace delete (#27175, @a-gajam)spans keyset in trace analytics migration 75868b020152 (#26710, @MatasVS)mlflow.openai.autolog() and other flavor autologging work on mlflow-skinny without numpy and pandas (#27038, @joshuawong-db)pyarrow 26 in core dependencies (#27216, @copilot-swe-agent)* as a registered model or prompt name when auth is enabled (#27202, @a-gajam)ConversationSimulator goal checks with Anthropic models that reject non-default temperature (#27170, @mlflow-app)timestamp=0 in mlflow.log_image and MlflowClient.log_image (#27179, @mlflow-app)IntegrityError (#27160, @a-gajam)gen_ai.tool.name for tool-call scoring of Google ADK and other GenAI semantic-convention traces (#27157, @mlflow-app)finish_reason in streamed OpenAI chat traces (#26973, @pcbeingused333)RetrievalRelevance fanout accounting in the scorer rate limiter (#26499, @james-fletcher-db)length and content_filter in AI Gateway (#27096, @mlflow-app)EvaluationDataset.merge_records() (#27119, @mlflow-app)FileStore.log_outputs (#27131, @joshuawong-db)mlflow.crewai spans (#27130, @joshuawong-db)mlflow.smolagents model spans (#27129, @joshuawong-db)upload_artifact_handler as the permission check (#27128, @joshuawong-db)Span.from_otel_proto, to_dict, from_dict and to_otel_proto (#26949, @selimaytac)MlflowClient.load_table() (#27057, @mlflow-app)content_filter in the gateway (#27018, @astroknot-sheep)log_table (#27061, @mlflow-app)extra_headers and headers LM kwargs from DSPy autologging (#27067, @mlflow-app)mlflow.artifactLocation trace tag (#27031, @joshuawong-db)workspaces/ artifact prefix (#27039, @joshuawong-db)is_unprotected_route (#27036, @joshuawong-db)FileStore (#26979, @tanghaoji)@scorer whose body is on the def line (#26972, @kriscon-db)SqlAlchemyStore (#26985, @its-amann)AnyType (#26998, @mlflow-app)mlflow server when its job runner fails (#26878, @kriscon-db)otel_api.py (#26940, @vatsalyd)server.json remotes during tool refresh (#26719, @dbasunag)attributes. prefix in search_experiments filter strings (#26923, @Blacksujit)mlflow.genai.evaluate cannot resolve mlflow.eval.requestId (#26769, @chris-chieu)/v1/traces (#26853, @mlflow-app)mlflow.login() and show the entered host in errors (#26886, @mlflow-app)_get_role_workspace_from_request (#26864, @tanghaoji)mlflow.pytorch (#26863, @tanghaoji)format_prompt replacing placeholders that appear inside variable values (#26807, @sayaksatpathi)--extract-fields output (#26824, @mlflow-app)mlflow.log_dict with mlflow.artifacts.load_dict (#26822, @mlflow-app)mlflow.genai.evaluate() produces no trace (#26767, @mlflow-app)search_traces ordering by end_time in the SQL store (#26761, @porkedchop)strands, semantic_kernel and haystack autolog exporting every span twice with a shared tracer provider (#26734, @LeviLong01)timestamp=0 in mlflow.log_metric and mlflow.log_metrics (#26765, @mlflow-app)AzureBlobArtifactRepository.create_multipart_upload to a fixed length (#26756, @github-actions)thoughts_token_count as output tokens in mlflow.gemini.autolog (#26744, @github-actions)None default for datetime ParamSpec in to_dict (#26729, @anishmehta24)EvaluationDataset.merge_records (#26730, @anishmehta24)useDatasetsPageQuery (#26721, @copilot-swe-agent)MLFLOW_DISABLE_TELEMETRY in the UI banner and settings (#26699, @copilot-swe-agent)mlflow.openai.autolog() tracing for streams closed early (#26667, @copilot-swe-agent)View All for time-filtered OSS traces (#26593, @B-Step62)re.fullmatch in name/ID validators to reject trailing newlines (#26485, @kayzenline)Array(datetime) prediction inputs (#26671, @copilot-swe-agent)properties being stripped from function tool schemas (#26646, @aaronteo-db)RegisterModelForm in dark mode (#26652, @github-actions)TokensCell hover card (#26642, @github-actions)gen_ai.system_instructions (#26470, @Zhuoxi2000)GeminiAdapter responses (#26446, @r14dd)ShowArtifactTextView when loading a new artifact (#26631, @github-actions)ShowArtifactAudioView (#26626, @github-actions)search_experiments for SQL stores (#26619, @github-actions)@mlflow.trace recording self in span inputs for falsy instances (#26617, @github-actions)api_base (#26594, @copilot-swe-agent)search_registered_models and search_model_versions (#26600, @copilot-swe-agent)TruLensScorer (#26574, @copilot-swe-agent)parallel_tool_calls translation for Anthropic and Vertex Claude (#26523, @copilot-swe-agent)DataFrame indexes (#26462, @aswanth-07)LogTracesDrawer dark mode (#26563, @copilot-swe-agent)timestamp (#26549, @copilot-swe-agent)--static-prefix in server job tracking URIs (#26546, @copilot-swe-agent)_create_table headers for empty CLI results (#26503, @BlueRaddish)dump_input_data serialization of dict inputs (#26519, @copilot-swe-agent)mlflow.claude_code (#26510, @B-Step62)search_logged_models status filters never matching on SQL store (#26512, @copilot-swe-agent)parse_outputs_to_str (#26451, @james-fletcher-db)delta.content in Databricks gateway streaming (#26439, @maharanay22)Documentation updates:
CLAUDE.md (#26957, @copilot-swe-agent)Small bug fixes and documentation updates:
#27428, #27352, #27351, #27278, @B-Step62; #27421, #27414, #27413, #27412, #27410, #27404, #27403, #27402, #27400, #27398, #27397, #27367, #27365, #27360, #27333, #27330, #27328, #27327, #27325, #27320, #27306, #27305, #27289, #27279, #27277, #27276, #27265, #27254, #27232, #27211, #27224, #27223, #27222, #27217, #27218, #27197, #27198, #27187, #27184, #27182, #27180, #27163, #27144, #27136, #27114, #27111, #27109, #27108, #27107, #27106, #27104, #27097, #27091, #27090, #27089, #27084, #27083, #27081, #27079, #27078, #27077, #27058, #27050, #27047, #27045, #27033, #27032, #27020, #27024, #27023, #27016, #26962, #26951, #26938, #26931, #26932, #26930, #26891, #26881, #26876, #26870, #26865, #26862, #26861, #26843, #26832, #26830, #26825, #26820, #26772, #26776, #26763, #26760, #26752, #26750, #26737, #26735, #26728, #26727, #26725, #26666, #26688, #26686, #26683, #26679, #26678, #26673, #26662, #26658, #26653, #26650, #26649, #26629, #26628, #26620, #26623, #26603, #26602, #26599, #26592, #26583, #26565, #26562, #26553, #26552, #26472, #26540, #26511, #26528, #26522, #26515, #26513, #26505, #26506, #26501, #26500, #26492, #26490, #26487, #26488, #26486, #26479, #26466, #26465, #26469, @copilot-swe-agent; #27399, #27312, #27293, #27302, #27246, #27235, #27069, #26867, #26960, #25307, #26718, #26708, #25851, #25847, #26701, #25844, #25843, #25842, #25841, #26588, #26589, @kriscon-db; #27376, @erdogan-deniz; #27394, #27393, #27350, #27326, #27113, @mlflow-app; #27377, #26748, @dependabot; #27364, #27366, #27259, #27313, #27041, #26981, #27017, #26983, #26906, #26885, #26514, #26476, #26467, @tanghaoji; #27165, #27040, #27148, #26912, #27034, #27029, #26993, #26952, #26789, #26920, #26779, #26707, #26461, @joshuawong-db; #27298, @v0ropaev; #26894, @Aditya-kumar-yadav; #27296, @davanstrien; #26090, @dbasunag; #27105, @Sonica-B; #27149, @Avicennasis; #26634, #26613, @shaikn6; #27135, #27143, #27021, #27013, #26953, #26905, #26796, #26598, #26541, #26352, @harupy; #27159, @smoorjani; #26901, #27027, #26768, @mprahl; #26946, @nananosirova; #26929, @Blacksujit; #26860, @alleyxd; #26850, #26632, @aaronteo-db; #26694, @Dedalo314
Your coding agent can read these notes before it upgrades. Set up the MCP server →
MLflow 3.17.2 is a patch release that contains a minor feature enhancement and bug fixes.
MLflow 3.17.2 is a patch release that contains a minor feature enhancement and bug fixes.
Features:
mlflow.genai.evaluate runs (#26543, @joshuawong-db)Bug fixes:
MlflowSparkStudy executors (#26714, @iyashk)MlflowStorage marking Optuna trials finished before their values are saved (#26713, @iyashk)MLflow 3.17.1 is a patch release that contains some minor feature enhancements, bug fixes, and documentation updates.
MLflow 3.17.1 is a patch release that contains some minor feature enhancements, bug fixes, and documentation updates.
Features:
make_judge() (#26578, @B-Step62)mlflow.genai.databricks.review_queues for managing review queues on Databricks (#26414, @harupy)Bug fixes:
Small bug fixes and documentation updates:
#26680, @joshuawong-db; #26584, @copilot-swe-agent; #26705, @kriscon-db
MLflow 3.17.0 includes several major features and improvements.
MLflow 3.17.0 includes several major features and improvements.
Database upgrade: SQL-backed servers require a coordinated schema upgrade with all writers stopped and a restorable backup, even when daily rollups remain disabled. Run mlflow db upgrade <database-url> before restarting all replicas on the new MLflow version. Mixed-version rolling upgrades are not supported. Follow the database upgrade instructions.
base_url for direct TypeSafe judges (#26445, @B-Step62)typesafe_api_base for TypeSafe Gateway endpoints (#26442, @B-Step62)name IN (...) filters for evaluation-dataset search (#26374, @HumairAK)@mlflow/typesafe tracing integration for TypeSafe AI (#26342, @B-Step62)TraceImage to custom trace views (#26297, @chris-chieu)setup.sh (#26341, @B-Step62)uc_model_service UC model-version dependency support (#26140, @TomeHirata)DatabricksUCModelService resource type (#26137, @TomeHirata)order_by in search_prompts() across the client, fluent, and GenAI APIs (#25967, @avanish-garg)AmazonBedrockProvider (#25863, @rhamzatov)createPresignedUploadUrl endpoint (#24765, @henishborad)mlflow.restore_experiment() and mlflow.restore_run() to the fluent API (#25852, @avanish-garg)charts: Expose Helm liveness/readiness/startup probes as configuration variables (#25560, @creater29)MLFLOW_ENABLE_AI_GATEWAY (#25236, @DaoDaoNoCode)@mlflow.trace and mlflow.start_span (#25548, @Rohitkanithi)Bug fixes:
EvaluationDataset hashing for list-valued targets and predictions (#26433, @LE0-Lin)version filters rewriting quoted values (#26423, @Amazinghorseli)response key for tool_call_response parts in GenAI semconv export (#26360, @Zhuoxi2000)MLFLOW_ALLOW_PICKLE_DESERIALIZATION guard to mlflow.paddle flavor (#26428, @aaronteo-db)copy.deepcopy and pickle on Span, LazySpan, and Trace (#26307, @Rohitkanithi)tool_call_id in GeminiAdapter (#26253, @codechrl)execute_tool and invoke_agent spans (#26394, @chrikrah)finish_reason tool_calls for Anthropic and Gemini tool calls (#26390, @r14dd)# hash route (#26378, @aaronteo-db)_strip_markdown_code_blocks for single-line and glued closing code fences (#26332, @lets-order-some-fries)server_name filters for MCP access endpoints (#26327, @Amazinghorseli)invoke_model_with_response_stream chunks and GPT-6 redactedContent reasoning (#26346, @kimnamu)@latest alias not resolving to the highest model version (#26291, @CTWalk)MlflowStorage.set_trial_state_values WAITING trial transitions (#25945, @CoralGarden52)tool_choice in Gemini and Vertex AI gateway requests (#26300, @Vaishnavi220506)MLFLOW_S3_IGNORE_TLS in multipart artifact download (#26004, @Anurag-M1)load_text and load_dict artifacts as UTF-8 (#26265, @ppcvote)_merge with AnyType order-independent (#26260, @JoeyTan21)llama_index traces during deferred streaming (#26293, @B-Step62)ERROR status when the stream raises mid-iteration (#26305, @iamhuman-cheolheelee)db extra (#26271, @abligail)db extra for SQLAlchemy 2.1 compatibility (#26239, @kriscon-db)like, ilike and in in search_logged_models filters (#26249, @SajalDevX)trace.text search missing non-ASCII span content (#26130, @pablo-legeren)GeminiAdapter (#26001, @rhamzatov)--trace-ids entries in mlflow traces delete (#26204, @breken-ai)search_logged_models pages (#26057, @lazarevtill)functionCall parts in GeminiAdapter (#26109, @r14dd)--order-by entries in mlflow traces search (#26192, @breken-ai)SearchUtils (#26180, @B-Step62)calculate_trace_filter_correlation binds integer experiment IDs (#26179, @B-Step62)mistral and groq autologging (#26177, @B-Step62)search_logged_models (#25983, @lazarevtill)trace_id property and fix __repr__ on Trace (#26007, @Rohitkanithi)@scorer functions whose signature contains parentheses (#26172, @anishmehta24)record_exception is given a string (#26171, @anishmehta24)ndcg_at_k document ID collisions (#26041, @kadyrbekovhamit-cyber)CreateModelVersion sources against declared resources (#26037, @joshuawong-db)navigator.onLine is false (#25961, @nrhys2005)responses model mode in the catalog sync (#26095, @TomeHirata)dataclass signatures (#26071, @HaokaiDing)server-info request instead of hanging the UI (#26027, @nrhys2005)send_stream_request (#26036, @rhamzatov)DefaultAzureCredential (#26031, @joshuawong-db)mlflow.types.responses (Note truncated.
MLflow 3.16.1 is a patch release that includes bug fixes and documentation updates.
MLflow 3.16.1 is a patch release that includes bug fixes and documentation updates.
Breaking changes:
basic_auth.ini (GHSA-gq3w-7jj3-x7gr) (#25751, @tanghaoji)Features:
timeout option to the @scorer decorator (#25720, @smoorjani)Bug fixes:
[UI] Deprecate the experiment Sessions tab in favor of Traces session grouping ( #25350 , @joshuawong-db )
MLflow 3.16.0 includes several major features and improvements
Breaking changes:
--static-prefix (#24511, @SeiichiroYoshioka)Features:
Expectation field to the traces table filter UI (#25460, @creater29)logFeedback and typed trace assessments (#25169, @rajudandigam)import mlflow (#25247, @harupy)shouldEnableModelTraceExplorerCustomTraceView in OSS (#25245, @aaronteo-db)mlflow scorers list --builtin with the full scorer catalog (#25003, @joshuawong-db)MLFLOW_TRACKING_AUTH=kubernetes/kubernetes-namespaced support (#24243, @robinnarsinghranabhat)optimize_prompts (#23488, @gaurav0107)--artifact-policy retarget to mlflow db move-resources for repointing experiment artifact roots (#24635, @gauthierpiarrette)generate_rationale_first through make_judge (#25002, @aaronteo-db)render_custom_view in Assistant (#24796, @aaronteo-db)Bug fixes:
update_webhook calls (#25509, @CoralGarden52)DatasetRecordSourceType.UNSPECIFIED dataset record sources (#25230, @2sumtech)IndexError in merge_records after to_dict/to_df on empty EvaluationDataset (#25498, @TomeHirata)update_webhook calls (#24842, @abligail)search_logged_models quoted-value parsing and invalid-operator message (#25398, @VenishPaneliya)@mlflow/openai Chat Completion spans open during streaming (#25416, @copilot-swe-agent)MLFLOW_ENABLE_REMOTE_ASSISTANT=true (#25394, @TomeHirata)traceId query param in trace navigation links (#25375, @NJAHNAVI2907)(name, options) provider entries (#25310, @rrioh)SpanBatcher from the trace flush paths so UC-table spans are exported (#25260, @alekseevpavel04)RunsChartsConfigureModal (#25283, @DaoDaoNoCode)openai autolog (#24618, @adamgurary)entry_point.getLogger() when unavailable in dynamic token config provider (#25207, @serena-ruan)log_spans() from the async event loop (#25185, @HumairAK)configure_autologging_for_evaluation configuring the wrong flavor (#25135, @Koustav-github)OpenInferenceTranslator (#25039, @felix-xsenda)SearchUtils._get_value from silently dropping dataset IN clause values (#25045, @kaikaila)python:3.10-slim-bullseye to python:3.11-slim-trixie (#24961, @mille-printemps)torch_dtype deprecation warnings on load and save paths (#25017, @rrioh)log_batch duplicate-metric recovery in SqlAlchemyStore by batch size (#24683, Note truncated.
MLflow 3.15.2 is a patch release that includes several major features and improvements.
MLflow 3.15.2 is a patch release that includes several major features and improvements.
Features:
Bug fixes:
databricks.sdk in Databricks to avoid telemetry deadlock (#24841, @aaronteo-db)runs.status constraint metadata (#24890, @joshuawong-db)MLflow 3.15.1 is a patch release that includes bug fixes and documentation updates.
MLflow 3.15.1 is a patch release that includes bug fixes and documentation updates.
env_pack on ARM client images (#24762, @qyc)MLflow 3.15.1 is a patch release that includes bug fixes and documentation updates.
Bug fixes:
env_pack on ARM client images (#24762) (#24835, @qyc)Documentation updates:
One column per month.