cd ~/series/demystifying-ai
Grep vs Vector Search - 93.1% vs 76.7% accuracy in agent benchmarks
EP 02 arXiv Jul 8, 2026 5 min

Is Grep All You Need?

A research team tested grep vs vector search inside AI agent loops. Grep won across every harness-model pair. But the real finding? The harness matters more than your retrieval method.

Share:
// TL;DR

Inline grep beats vector retrieval in every agent harness tested. But switching harnesses caused a bigger accuracy swing (16 points) than switching retrieval methods. The architecture wrapping your search matters more than the search itself.

# The Experiment

116 questions from LongMemEval. Four agent harnesses. Two retrieval methods. The setup nobody had tested before.

Grep (Lexical)
Exact pattern matching
Zero hallucinated context
Instant, deterministic results
No GPU, no embeddings
Vector (Semantic)
~ Approximate matching
~ Can surface paraphrases
~ Semantic range = noise in tight loops
~ Requires embeddings + index

# The Numbers

93.1%
Best: Grep + Chronos
76.7%
Same model, diff harness
16pt
Gap from harness alone

The punchline: Switching from vector to grep improved accuracy. But switching harnesses with the same retrieval method caused an even bigger swing. The architecture wrapping your search is the real variable.

# Why Grep Wins in Agent Loops

Speed

Agent loops iterate fast. Grep returns in milliseconds. No embedding computation, no index lookup, no network round-trip.

🎯

Precision

When it hits, it hits exactly. No "close but not quite" results polluting the context window with noise.

💰

Token Efficiency

Grep returns only matching lines. Vector search returns full chunks with surrounding context you don't need.

🔬

Determinism

Same query, same results, every time. No embedding drift, no index staleness, no similarity threshold tuning.

# The Real Insight: Harness > Retrieval

The paper tested four harnesses with the same model and retrieval. The harness made the difference, not grep vs vector.

Chronos
93.1%
Codex
86.2%
Gemini
82.8%
Claude
76.7%

All using the same model (Claude Opus 4.6) + inline grep. The only variable is the agent harness. Dynamic prompts, tool-calling style, and how results are presented to the model created a 16-point accuracy gap.

# The Plot Twist: File-Based Results

When results were written to files instead of inlined into the context window, vector search came back. It beat grep in 5 out of 10 configurations.

Inline Results

Results appended directly into context window.

Grep wins. Precision matters when context is crowded.

File-Based Results

Results written to files, agent reads separately.

Vector competitive. Semantic range becomes an asset when agent reasons deliberately.

# When to Use What

Tight agent loops grep

Precision + speed + token efficiency

File-based results vector

Semantic range is an asset here

Known patterns (IDs, errors) grep

Exact match = zero noise

Conceptual questions vector

Needs semantic understanding

Production system harness

16pt gap from harness > retrieval

// Bottom Line

Don't default to vector search. Inside tight agent loops, grep is faster, cheaper, and more predictable. But the real lesson: stop obsessing over retrieval and start investing in your agent harness. The prompts, tool-calling patterns, and result delivery format matter more than whether you use grep or embeddings.

NEXT EPISODE
Saturday
#03 Upcoming

LongCat-2.0: Open-Source 1M Context That Beats GPT-5.5

Meituan just dropped a 1.6T parameter model with 1M token context. MIT licensed. Outperforms GPT-5.5 on SWE-bench Pro.

Enjoyed this?

New episodes Mon, Wed, Sat.