diff --git a/STATE.md b/STATE.md
index a81b2d9..847c9df 100644
--- a/STATE.md
+++ b/STATE.md
@@ -22,3 +22,4 @@
 - [2026-03-12] Harberger tax staking mechanism
 - [2026-03-13] LLM seed — Momentum Follower optimizer (#695)
 - [2026-03-14] evolve.sh auto-incrementing per-run results directory (#752)
+- [2026-03-14] EVAL_MODE now defaults to revm (#751)
diff --git a/tools/push3-evolution/evolve.sh b/tools/push3-evolution/evolve.sh
index ea0a96b..924e3f8 100755
--- a/tools/push3-evolution/evolve.sh
+++ b/tools/push3-evolution/evolve.sh
@@ -31,7 +31,8 @@
 # Algorithm:
 #   1. Initialize population: N copies of seed, each with M random mutations.
 #   2. For each generation:
-#      a. Score all candidates via fitness.sh
+#      a. Score all candidates via the configured fitness backend
+#         (batch-eval.sh for revm [default], fitness.sh for anvil)
 #      b. Log generation stats (min/max/mean fitness, best candidate)
 #      c. Select k survivors via tournament selection (k = population/2)
 #      d. Elitism: copy top N candidates unchanged into next generation
@@ -49,7 +50,8 @@
 #     evolution.log        full run log
 #
 # Environment:
-#   ANVIL_FORK_URL  Passed through to fitness.sh when Anvil is not running.
+#   BASE_RPC_URL    Required when EVAL_MODE=revm (the default). Base network RPC endpoint.
+#   ANVIL_FORK_URL  Passed through to fitness.sh when EVAL_MODE=anvil.
 #
 # TSX resolution order: tsx in PATH → node_modules/.bin/tsx → npx tsx.
 # =============================================================================
@@ -66,10 +68,10 @@ MUTATE_CLI="$SCRIPT_DIR/mutate-cli.ts"
 SEED_GEN_CLI="$SCRIPT_DIR/seed-gen-cli.ts"
 
 # EVAL_MODE controls which fitness backend is used:
-#   anvil  (default) — per-candidate Anvil+forge-script pipeline (fitness.sh)
-#   revm            — in-process revm via FitnessEvaluator.t.sol (batch-eval.sh)
-#                     Requires BASE_RPC_URL env var.  10-100× faster at scale.
-EVAL_MODE="${EVAL_MODE:-anvil}"
+#   revm   (default) — in-process revm via FitnessEvaluator.t.sol (batch-eval.sh)
+#                      Requires BASE_RPC_URL env var.  10-100× faster at scale.
+#   anvil            — per-candidate Anvil+forge-script pipeline (fitness.sh)
+EVAL_MODE="${EVAL_MODE:-revm}"
 
 # =============================================================================
 # Argument parsing