diff --git a/nlp-service/requirements.txt b/nlp-service/requirements.txt
index fcf249a..4c7e201 100644
--- a/nlp-service/requirements.txt
+++ b/nlp-service/requirements.txt
@@ -1,30 +1,30 @@
 # Kompl v2 nlp-service — commit 4. Pins per docs/research/2026-04-08-conversion-deps.md
 # and docs/research/2026-04-09-llm-compile.md.
-fastapi==0.139.0
-uvicorn[standard]==0.51.0
+fastapi==0.141.1
+uvicorn[standard]==0.52.3
 pydantic==2.13.4
 httpx==0.28.1
-markitdown[all]==0.1.6
+markitdown[all]==0.1.7
 # Promoted to a direct dep so routers/conversion.py can call
 # youtube-transcript-api WITHOUT going through MarkItDown for YouTube URLs.
 # Reason: MarkItDown's YouTubeConverter silently falls back to scraping the
 # watch-page HTML when no transcript exists (returns ~800 chars of footer
 # chrome — verified live on session 4a00f339). We need the transcript-vs-no-
 # transcript signal explicitly, which only the direct API exposes.
-# markitdown[all]==0.1.6 constrains ~=1.0.0; Dockerfile upgrades to 1.2.x
+# markitdown[all]==0.1.7 constrains ~=1.0.0; Dockerfile upgrades to 1.2.x
 # after pip install (1.0.x breaks on current YouTube timedtext responses).
 youtube-transcript-api~=1.0.0
 # Commit 4: LLM compile. Use google-genai NOT google-generativeai (deprecated 2025-Q1).
-google-genai==2.11.0
+google-genai==2.18.1
 # Commit 4: async token-bucket rate limiter. Single uvicorn worker only —
 # InMemoryBucket is process-local; see research artifact section 3.
 pyrate-limiter==3.9.0
 # Part 2a: multi-layer NLP extraction pipeline.
-spacy==3.8.14
+spacy==3.8.15
 rake-nltk==1.0.6
 yake==0.7.3
 keybert==0.9.0
-sentence-transformers==5.6.0
+sentence-transformers==5.7.0
 # Dockerfile pins torch==2.6.0 (CPU wheel). torch 2.5.1 breaks transformers
 # 5.x imports; torch >=2.7 breaks sentence-transformers meta-device init
 # (NotImplementedError on .to(device)).
@@ -35,7 +35,7 @@ transformers==5.9.0
 pytextrank==3.3.0
 scikit-learn==1.9.0
 # nltk: required by rake-nltk (stopwords + punkt tokenizer data).
-nltk==3.10.0
+nltk==3.10.3
 # Part 2b: fast Levenshtein/Jaro-Winkler matching for entity resolution.
 rapidfuzz>=3.14.5
 # Recovers truncated JSON from Gemini 2.5 Flash repetition-loop bug