fix: Double embedding pass on fresh index

README.md

@@ -24,7 +24,7 @@
 
 </div>
 
-Semble is a code search library built for agents. It returns the exact code snippets they need instantly, using ~99% fewer tokens than grep+read. Indexing and searching a full codebase end-to-end takes under a second, matching the retrieval quality of a code-specialized transformer while indexing ~220x faster and querying ~17x faster (see [benchmarks](#benchmarks)). Everything runs on CPU with no API keys, GPU, or external services. Use it as an MCP server, a CLI tool via AGENTS.md, or a dedicated sub-agent, and any coding agent (Claude Code, Cursor, Codex, OpenCode, etc.) gets instant access to any repo.
+Semble is a code search library built for agents. It returns the exact code snippets they need instantly, using ~99% fewer tokens than grep+read. Indexing and searching a full codebase end-to-end takes under a second, matching the retrieval quality of a code-specialized transformer while indexing ~340x faster and querying ~17x faster (see [benchmarks](#benchmarks)). Everything runs on CPU with no API keys, GPU, or external services. Use it as an MCP server, a CLI tool via AGENTS.md, or a dedicated sub-agent, and any coding agent (Claude Code, Cursor, Codex, OpenCode, etc.) gets instant access to any repo.
 
 ## Quickstart
 
@@ -244,7 +244,7 @@ We benchmark quality and speed across ~1,250 queries over 63 repositories in 19
 </tr>
 </table>
 
-The quality benchmark (left) scores retrieval quality (NDCG@10) against total latency; semble matches the quality of the 137M-parameter [CodeRankEmbed](https://huggingface.co/nomic-ai/CodeRankEmbed) while indexing 220x faster. The token efficiency benchmark (right) measures how many tokens each method needs to reach a given recall level; semble uses 99% fewer tokens on average and hits 97% recall at only 2k tokens, while grep+read needs a full 100k context window to reach 85%. See [benchmarks](benchmarks/README.md) for per-language results, ablations, and full methodology.
+The quality benchmark (left) scores retrieval quality (NDCG@10) against total latency; semble matches the quality of the 137M-parameter [CodeRankEmbed](https://huggingface.co/nomic-ai/CodeRankEmbed) while indexing 340x faster. The token efficiency benchmark (right) measures how many tokens each method needs to reach a given recall level; semble uses 99% fewer tokens on average and hits 97% recall at only 2k tokens, while grep+read needs a full 100k context window to reach 85%. See [benchmarks](benchmarks/README.md) for per-language results, ablations, and full methodology.
 
 ## How it works
 

benchmarks/README.md

@@ -17,7 +17,7 @@ Quality and speed across all methods.
 
 | Method               |   NDCG@10 |      Index |   Query p50 |
 | -------------------- | --------: | ---------: | ----------: |
-| **semble**           | **0.854** | **518 ms** | **0.91 ms** |
+| **semble**           | **0.854** | **344 ms** | **0.91 ms** |
 | CodeRankEmbed        |     0.839 |      116 s |       16 ms |
 | ColGREP              |     0.693 |      5.4 s |      122 ms |
 | BM25                 |     0.673 |      47 ms |     0.17 ms |
@@ -32,7 +32,7 @@ Quality and speed across all methods.
 | :-----------------------------------------------------------------: | :-----------------------------------------------------------------: |
 |          _Time to first result (index + query) vs NDCG@10_          |             _Query latency on a warm index vs NDCG@10_              |
 
-semble matches the NDCG@10 of the 137M-param CodeRankEmbed while winning index time by ~220x and query latency by ~17x.
+semble matches the NDCG@10 of the 137M-param CodeRankEmbed while winning index time by ~340x and query latency by ~17x.
 
 NDCG@10 is averaged across all queries. Speed numbers use one repo per language, CPU only: cold-start index time and warm query p50 (median across 5 consecutive runs).
 

benchmarks/plot.py

@@ -96,7 +96,7 @@ class _Method(TypedDict):
     {
         "name": "semble",
         "ndcg10": 0.8544,
-        "index_ms": 518.2,
+        "index_ms": 343.5,
         "query_p50_ms": 0.91,
         "color": "#1a5fa8",
         "params_m": 16,