diff --git a/src/modules/semantle/README.md b/src/modules/semantle/README.md
index 9755494..f6b8eca 100644
--- a/src/modules/semantle/README.md
+++ b/src/modules/semantle/README.md
@@ -28,11 +28,12 @@ the target pool and the vocabulary** — so every legal guess could itself
 have been the answer, and OOV detection is an O(1) `Set.has()` with no
 upstream round-trip. Regenerate with `node scripts/build-semantle-words.js`.
 
-**Similarity:** `@cf/baai/bge-small-en-v1.5` text embeddings via the
+**Similarity:** `@cf/baai/bge-m3` multilingual text embeddings via the
 `env.AI` binding. Each in-vocab guess runs one inference call batching
-target + guess (384-dim vectors) and the module scores them with local
-cosine similarity. At ~0.0037 Neurons per guess, the Workers Free plan
-cap of 10k Neurons/day covers ~2.7M guesses/day.
+target + guess (1024-dim vectors) and the module scores them with local
+cosine similarity. At 1075 Neurons per M input tokens (~0.002 N/guess
+for short words), the Workers Free plan cap of 10k Neurons/day covers
+~4.6M guesses/day. Same model as `doantu` so both share the binding.
 
 OOV guesses short-circuit before inference — the player sees
 "isn't in the vocabulary" instead of a noisy subword-based score.
@@ -70,8 +71,8 @@ Each `guesses[]` entry is `{ word, canonical, similarity }`.
 
 ## Config
 
-No env vars. Model defaults to `@cf/baai/bge-small-en-v1.5`; override with
-`createClient(env.AI, { model: "@cf/baai/bge-base-en-v1.5" })` in a test
+No env vars. Model defaults to `@cf/baai/bge-m3`; override with
+`createClient(env.AI, { model: "@cf/baai/bge-small-en-v1.5" })` in a test
 or alternative deploy.
 
 ## Why unlimited guesses?
@@ -82,6 +83,6 @@ tracked via `bestGuessCount` — fewest guesses to solve across all rounds.
 
 ## Credits
 
-- Embeddings: [`@cf/baai/bge-small-en-v1.5`](https://developers.cloudflare.com/workers-ai/models/bge-small-en-v1.5/) on Cloudflare Workers AI.
+- Embeddings: [`@cf/baai/bge-m3`](https://developers.cloudflare.com/workers-ai/models/bge-m3/) on Cloudflare Workers AI (multilingual).
 - Target dictionary: [google-10000-english](https://github.com/first20hours/google-10000-english) by Josh Kaufman, derived from Peter Norvig's Google Ngram analysis.
 - Game concept: [Semantle](https://semantle.com/) by David Turner.
diff --git a/src/modules/semantle/api-client.js b/src/modules/semantle/api-client.js
index 32cc80b..8682f11 100644
--- a/src/modules/semantle/api-client.js
+++ b/src/modules/semantle/api-client.js
@@ -1,9 +1,9 @@
 /**
  * @file Cloudflare Workers AI client for the semantle module.
  *
- * Runs the `@cf/baai/bge-small-en-v1.5` text-embedding model via the `env.AI`
- * binding, then scores guesses by computing cosine similarity between the
- * target and guess vectors locally (no extra round-trip).
+ * Runs the `@cf/baai/bge-m3` text-embedding model via the `env.AI` binding,
+ * then scores guesses by computing cosine similarity between the target
+ * and guess vectors locally (no extra round-trip).
  *
  * Vocabulary: the curated `words-data.js` list (google-10k) doubles as our
  * in/out-of-vocabulary set — anything outside it is treated as OOV so players
@@ -13,7 +13,9 @@
 import { randomLine } from "./wordlist.js";
 import WORDS from "./words-data.js";
 
-const DEFAULT_MODEL = "@cf/baai/bge-small-en-v1.5";
+// BGE-M3: multilingual, 1024 dimensions, 1075 Neurons per M input tokens —
+// cheaper than bge-small-en-v1.5 (1841 N/M) and matches the doantu sibling.
+const DEFAULT_MODEL = "@cf/baai/bge-m3";
 
 // O(1) membership lookup for OOV detection. Built once per isolate.
 const VOCAB = new Set(WORDS);
diff --git a/src/modules/semantle/index.js b/src/modules/semantle/index.js
index c0508fb..3349796 100644
--- a/src/modules/semantle/index.js
+++ b/src/modules/semantle/index.js
@@ -4,7 +4,7 @@
  * Targets come from a curated local wordlist (same list doubles as the
  * vocabulary for OOV detection, so no upstream check is needed to pick or
  * validate a word). Similarity scores come from cosine distance between
- * `@cf/baai/bge-small-en-v1.5` embeddings produced by the `env.AI` binding.
+ * `@cf/baai/bge-m3` multilingual embeddings produced by the `env.AI` binding.
  */
 
 import { createClient } from "./api-client.js";
diff --git a/tests/modules/semantle/api-client.test.js b/tests/modules/semantle/api-client.test.js
index 9acc464..a429846 100644
--- a/tests/modules/semantle/api-client.test.js
+++ b/tests/modules/semantle/api-client.test.js
@@ -2,10 +2,10 @@ import { describe, expect, it, vi } from "vitest";
 import { UpstreamError, createClient } from "../../../src/modules/semantle/api-client.js";
 
 /**
- * Build a deterministic 384-dim vector (bge-small output size) from a seed
- * so cosine scores are reproducible without hardcoding 384 floats.
+ * Build a deterministic 1024-dim vector (bge-m3 output size) from a seed
+ * so cosine scores are reproducible without hardcoding 1024 floats.
  */
-function fakeVector(seed, dim = 384) {
+function fakeVector(seed, dim = 1024) {
   const out = new Array(dim);
   for (let i = 0; i < dim; i++) out[i] = Math.sin(seed * (i + 1));
   return out;
@@ -45,14 +45,14 @@ describe("semantle/api-client", () => {
 
     it("similarity batches target + guess in a single run() call", async () => {
       const ai = fakeAi(async (_model, { text }) => ({
-        shape: [text.length, 384],
+        shape: [text.length, 1024],
         data: text.map((_, i) => fakeVector(i + 1)),
       }));
       const client = createClient(ai);
       await client.similarity("apple", "orange");
       expect(ai.run).toHaveBeenCalledTimes(1);
       const [model, input] = ai.run.mock.calls[0];
-      expect(model).toBe("@cf/baai/bge-small-en-v1.5");
+      expect(model).toBe("@cf/baai/bge-m3");
       expect(input).toEqual({ text: ["apple", "orange"] });
     });
 
@@ -107,7 +107,7 @@ describe("semantle/api-client", () => {
     });
 
     it("similarity returns null score when a vector norm is zero", async () => {
-      const zero = new Array(384).fill(0);
+      const zero = new Array(1024).fill(0);
       const ai = fakeAi(async () => ({ data: [zero, fakeVector(1)] }));
       const client = createClient(ai);
       const res = await client.similarity("apple", "orange");
diff --git a/wrangler.toml b/wrangler.toml
index 21b2c5d..5c44305 100644
--- a/wrangler.toml
+++ b/wrangler.toml
@@ -25,11 +25,12 @@ binding = "DB"
 database_name = "miti99bot-db"
 database_id = "261b54e7-0fdb-4fe7-8ed9-2e8a8bcf459c"
 
-# Workers AI — inference binding used by the semantle module for
-# @cf/baai/bge-small-en-v1.5 text embeddings (replaces ConceptNet upstream).
-# Accessed as `env.AI` in handlers. Included on the Workers Free plan:
-# 10,000 Neurons/day at no charge (hard-stops — no billing on Free plan).
-# bge-small is ~0.0037 Neurons/guess → ~2.7M guesses/day within the cap.
+# Workers AI — inference binding used by semantle + doantu for
+# @cf/baai/bge-m3 multilingual text embeddings. Accessed as `env.AI`
+# in handlers. Included on the Workers Free plan: 10,000 Neurons/day at
+# no charge (hard-stops — no billing on Free plan).
+# bge-m3 is 1075 Neurons per M input tokens → ~0.002 N/guess (2 short
+# words), ~4.6M guesses/day within the cap.
 # Pricing: https://developers.cloudflare.com/workers-ai/platform/pricing/
 [ai]
 binding = "AI"