LLM observability SDK downloads, 2026-09

Downloads of each listed tool's headline PyPI and npm package from 2026-09-01 to 2026-09-30, compared with 2026-08. Measured 2026-10-07; run status complete. The largest PyPI mover with at least 1,000 downloads in 2026-08 was Lunary (+147.9%).

Package downloads include CI runs and installs pulled in as dependencies; they are not people or customers. PyPI and npm are never added together. Movers need at least 1,000 downloads in the previous month.

Top movers by category

Tracing & observability

Lunary (PyPI lunary): 61,206 → 151,702 (+147.9%)
Last9 (AI / LLM observability) (PyPI last9-genai): 1,641 → 2,419 (+47.4%)
Galtea (PyPI galtea): 6,462 → 8,600 (+33.1%)
Pydantic Logfire (npm @pydantic/logfire-node): 143,999 → 362,350 (+151.6%)
Laminar (npm @lmnr-ai/lmnr): 149,237 → 214,156 (+43.5%)
Langfuse (npm @langfuse/tracing): 7,272,141 → 9,943,547 (+36.7%)

Evaluation

Lunary (PyPI lunary): 61,206 → 151,702 (+147.9%)
Galtea (PyPI galtea): 6,462 → 8,600 (+33.1%)
Okareo (PyPI okareo): 6,955 → 8,726 (+25.5%)
Pydantic Logfire (npm @pydantic/logfire-node): 143,999 → 362,350 (+151.6%)
Laminar (npm @lmnr-ai/lmnr): 149,237 → 214,156 (+43.5%)
Langfuse (npm @langfuse/tracing): 7,272,141 → 9,943,547 (+36.7%)

Guardrails & safety

NeMo Guardrails (PyPI nemoguardrails): 400,715 → 494,127 (+23.3%)
Fiddler AI (PyPI fiddler-otel): 2,651 → 3,256 (+22.8%)
Portkey (PyPI portkey-ai): 3,976,719 → 4,190,600 (+5.4%)
OpenLIT (npm openlit): 4,112 → 5,398 (+31.3%)
Orq.ai (npm @orq-ai/node): 170,560 → 198,137 (+16.2%)
Promptfoo (npm promptfoo): 2,700,713 → 3,026,104 (+12.0%)

Prompt management

Lunary (PyPI lunary): 61,206 → 151,702 (+147.9%)
Portkey (PyPI portkey-ai): 3,976,719 → 4,190,600 (+5.4%)
Braintrust (PyPI braintrust): 6,891,803 → 6,829,474 (-0.9%)
Langfuse (npm @langfuse/tracing): 7,272,141 → 9,943,547 (+36.7%)
OpenLIT (npm openlit): 4,112 → 5,398 (+31.3%)
Braintrust (npm braintrust): 5,937,478 → 7,104,136 (+19.6%)

Gateways with observability

Portkey (PyPI portkey-ai): 3,976,719 → 4,190,600 (+5.4%)
Pydantic Logfire (PyPI logfire): 13,707,474 → 13,302,869 (-3.0%)
Helicone (PyPI helicone-helpers): 20,778 → 18,405 (-11.4%)
Pydantic Logfire (npm @pydantic/logfire-node): 143,999 → 362,350 (+151.6%)
Orq.ai (npm @orq-ai/node): 170,560 → 198,137 (+16.2%)
Helicone (npm @helicone/helpers): 14,519 → 16,327 (+12.5%)

PyPI, all headline packages

PyPI downloads per headline package, 2026-09
Tool	Package	This month	Previous month	Change	Reproduce
LiteLLM	PyPI`litellm`	91,220,891	639,210,245	-85.7%	query
LangSmith	PyPI`langsmith`	79,059,601	114,398,334	-30.9%	query
Langfuse	PyPI`langfuse`	23,586,307	26,123,802	-9.7%	query
MLflow	PyPI`mlflow-tracing`	15,557,546	24,610,364	-36.8%	query
Pydantic Logfire	PyPI`logfire`	13,302,869	13,707,474	-3.0%	query
Pydantic Evals	PyPI`pydantic-evals`	8,186,335	11,564,368	-29.2%	query
Braintrust	PyPI`braintrust`	6,829,474	6,891,803	-0.9%	query
Portkey	PyPI`portkey-ai`	4,190,600	3,976,719	+5.4%	query
Inspect AI	PyPI`inspect-ai`	2,801,769	7,917,325	-64.6%	query
DeepEval / Confident AI	PyPI`deepeval`	2,601,822	4,770,912	-45.5%	query
Traceloop / OpenLLMetry	PyPI`traceloop-sdk`	1,808,065	2,066,920	-12.5%	query
Opik	PyPI`opik`	1,784,146	2,975,474	-40.0%	query
lm-evaluation-harness	PyPI`lm-eval`	1,059,656	1,711,783	-38.1%	query
Ragas	PyPI`ragas`	1,049,561	1,498,206	-29.9%	query
Evidently AI	PyPI`evidently`	858,288	1,297,000	-33.8%	query
W&B Weave	PyPI`weave`	815,986	996,510	-18.1%	query
Arize Phoenix	PyPI`arize-phoenix`	685,423	2,099,334	-67.4%	query
NeMo Guardrails	PyPI`nemoguardrails`	494,127	400,715	+23.3%	query
Arize AX	PyPI`arize-otel`	351,534	359,307	-2.2%	query
OpenLIT	PyPI`openlit`	283,972	417,431	-32.0%	query
Openlayer	PyPI`openlayer`	151,776	158,179	-4.0%	query
Lunary	PyPI`lunary`	151,702	61,206	+147.9%	query
Azure AI Evaluation SDK (Microsoft Foundry)	PyPI`azure-ai-evaluation`	137,960	257,525	-46.4%	query
AgentOps	PyPI`agentops`	109,821	244,811	-55.1%	query
Orq.ai	PyPI`orq-ai-sdk`	109,001	124,947	-12.8%	query
PromptLayer	PyPI`promptlayer`	100,379	161,270	-37.8%	query
Judgment Labs (Judgeval)	PyPI`judgeval`	61,808	82,190	-24.8%	query
garak	PyPI`garak`	40,845	69,566	-41.3%	query
Agenta	PyPI`agenta`	38,943	58,084	-33.0%	query
Galileo	PyPI`galileo`	38,109	39,437	-3.4%	query
TruLens	PyPI`trulens`	23,807	37,951	-37.3%	query
Helicone	PyPI`helicone-helpers`	18,405	20,778	-11.4%	query
Giskard	PyPI`giskard`	16,784	31,201	-46.2%	query
Promptfoo	PyPI`promptfoo`	12,747	15,740	-19.0%	query
Latitude	PyPI`latitude-telemetry`	10,761	16,861	-36.2%	query
HoneyHive	PyPI`honeyhive`	10,548	16,144	-34.7%	query
Okareo	PyPI`okareo`	8,726	6,955	+25.5%	query
Galtea	PyPI`galtea`	8,600	6,462	+33.1%	query
LlamaFirewall	PyPI`llamafirewall`	4,845	4,913	-1.4%	query
Fiddler AI	PyPI`fiddler-otel`	3,256	2,651	+22.8%	query
Last9 (AI / LLM observability)	PyPI`last9-genai`	2,419	1,641	+47.4%	query
Scorecard	PyPI`scorecard-ai`	1,590	1,787	-11.0%	query
UpTrain	PyPI`uptrain`	1,270	2,023	-37.2%	query
AI Verify Moonshot	PyPI`aiverify-moonshot`	1,088	2,482	-56.2%	query
Scorable (formerly Root Signals)	PyPI`scorable`	340	443	-23.3%	query

npm, all headline packages

npm downloads per headline package, 2026-09
Tool	Package	This month	Previous month	Change	Reproduce
LangSmith	npm`langsmith`	27,052,306	26,562,510	+1.8%	query
Langfuse	npm`@langfuse/tracing`	9,943,547	7,272,141	+36.7%	query
Braintrust	npm`braintrust`	7,104,136	5,937,478	+19.6%	query
Promptfoo	npm`promptfoo`	3,026,104	2,700,713	+12.0%	query
W&B Weave	npm`weave`	2,030,408	1,980,353	+2.5%	query
Portkey	npm`portkey-ai`	803,024	899,261	-10.7%	query
OpenEvals	npm`openevals`	545,638	538,123	+1.4%	query
Traceloop / OpenLLMetry	npm`@traceloop/node-server-sdk`	519,626	653,998	-20.5%	query
Raindrop	npm`raindrop-ai`	374,634	627,329	-40.3%	query
Pydantic Logfire	npm`@pydantic/logfire-node`	362,350	143,999	+151.6%	query
Arize Phoenix	npm`@arizeai/phoenix-otel`	342,396	313,321	+9.3%	query
Laminar	npm`@lmnr-ai/lmnr`	214,156	149,237	+43.5%	query
Orq.ai	npm`@orq-ai/node`	198,137	170,560	+16.2%	query
LangWatch	npm`langwatch`	191,292	228,236	-16.2%	query
Opik	npm`opik`	160,297	163,875	-2.2%	query
Judgment Labs (Judgeval)	npm`judgeval`	143,216	122,062	+17.3%	query
Latitude	npm`@latitude-data/telemetry`	23,053	28,214	-18.3%	query
PromptLayer	npm`promptlayer`	20,754	27,061	-23.3%	query
Helicone	npm`@helicone/helpers`	16,327	14,519	+12.5%	query
Scorecard	npm`scorecard-ai`	8,277	8,487	-2.5%	query
Galileo	npm`galileo`	5,803	6,909	-16.0%	query
OpenLIT	npm`openlit`	5,398	4,112	+31.3%	query
Openlayer	npm`openlayer`	1,279	820	+56.0%	query
Respan	npm`@respan/respan`	470	904	-48.0%	query
AgentOps	npm`agentops`	365	6,343	-94.2%	query
HoneyHive	npm`@honeyhive/api-client`	308	258	+19.4%	query
Okareo	npm`okareo-ts-sdk`	151	228	-33.8%	query
Future AGI	npm`@future-agi/ai-evaluation`	59	960	-93.9%	query

Data file: data/reports/2026-09.json. Method: methodology.