Adding Voyage v4 models

docs/vectorizer/quick-start-voyage.md

@@ -102,8 +102,17 @@ Now you can create and run a vectorizer. A vectorizer is a pgai concept, it proc
     ```
 
     **Available Voyage AI Models:**
-    - `voyage-3.5-lite`: Cost & latency optimized, 1024 dims (1M tokens/request) - **Recommended**
+
+    **Voyage 4 Series (Latest):**
+    - `voyage-4-lite`: Cost & latency optimized, 1024 dims (1M tokens/request) - **Recommended**
+    - `voyage-4`: General-purpose optimized, 1024 dims (320K tokens/request)
+    - `voyage-4-large`: Best retrieval quality, 1024 dims (120K tokens/request)
+
+    **Voyage 3.5 Series:**
+    - `voyage-3.5-lite`: Cost & latency optimized, 1024 dims (1M tokens/request)
     - `voyage-3.5`: General-purpose optimized, 1024 dims (320K tokens/request)
+
+    **Voyage 3 Series:**
     - `voyage-3-large`: Best for general-purpose & multilingual, 1024 dims (120K tokens/request)
     - `voyage-code-3`: Specialized for code retrieval, 1024 dims (120K tokens/request)
     - `voyage-finance-2`: Finance domain optimized, 1024 dims

projects/extension/docs/model_calling/voyageai.md

@@ -31,7 +31,7 @@ as a [session level parameter]. For more options and details, consult the
    `ai.voyage_api_key` is set for the duration of your psql session, you do not need to specify it for pgai functions.
 
     ```sql
-    SELECT * FROM ai.voyageai_embed('voyage-3-lite', 'sample text to embed');
+    SELECT * FROM ai.voyageai_embed('voyage-4-lite', 'sample text to embed');
     ```
 
 ## Usage
@@ -49,7 +49,7 @@ Generate [embeddings](https://docs.voyageai.com/docs/embeddings) using a specifi
 
     ```sql
     SELECT ai.voyageai_embed
-    ( 'voyage-3-lite'
+    ( 'voyage-4-lite'
     , 'the purple elephant sits on a red mushroom'
     );
     ```
@@ -67,7 +67,7 @@ Generate [embeddings](https://docs.voyageai.com/docs/embeddings) using a specifi
 
     ```sql
     SELECT ai.voyageai_embed
-    ( 'voyage-3-lite'
+    ( 'voyage-4-lite'
     , array['Timescale is Postgres made Powerful', 'the purple elephant sits on a red mushroom']
     );
     ```
@@ -80,7 +80,7 @@ Generate [embeddings](https://docs.voyageai.com/docs/embeddings) using a specifi
 
     ```sql
     SELECT ai.voyageai_embed
-    ( 'voyage-3-lite'
+    ( 'voyage-4-lite'
     , 'A query'
     , input_type => 'query'
     );

projects/extension/tests/test_voyageai.py

@@ -143,6 +143,63 @@ def test_voyageai_embed_with_multiple_inputs(cur, voyageai_api_key):
     assert result == 2
 
 
+def test_voyageai_embed_voyage_4(cur, voyageai_api_key):
+    """Test voyage-4 model."""
+    cur.execute(
+        """
+        select vector_dims
+        (
+            ai.voyageai_embed
+            ( 'voyage-4'
+            , 'hello world'
+            , api_key=>%s
+            )
+        )
+    """,
+        (voyageai_api_key,),
+    )
+    actual = cur.fetchone()[0]
+    assert actual == 1024
+
+
+def test_voyageai_embed_voyage_4_lite(cur, voyageai_api_key):
+    """Test voyage-4-lite model."""
+    cur.execute(
+        """
+        select vector_dims
+        (
+            ai.voyageai_embed
+            ( 'voyage-4-lite'
+            , 'hello world'
+            , api_key=>%s
+            )
+        )
+    """,
+        (voyageai_api_key,),
+    )
+    actual = cur.fetchone()[0]
+    assert actual == 1024
+
+
+def test_voyageai_embed_voyage_4_large(cur, voyageai_api_key):
+    """Test voyage-4-large model."""
+    cur.execute(
+        """
+        select vector_dims
+        (
+            ai.voyageai_embed
+            ( 'voyage-4-large'
+            , 'hello world'
+            , api_key=>%s
+            )
+        )
+    """,
+        (voyageai_api_key,),
+    )
+    actual = cur.fetchone()[0]
+    assert actual == 1024
+
+
 def test_voyageai_embed_voyage_3_5_lite(cur, voyageai_api_key):
     """Test voyage-3.5-lite model (current naming)."""
     cur.execute(

projects/pgai/pgai/vectorizer/embedders/voyageai.py

@@ -18,13 +18,13 @@
 def voyage_max_tokens_per_batch(model: str) -> int:
     # According to https://docs.voyageai.com/docs/embeddings:
     # The total number of tokens in the list is at most:
-    # - 1M for voyage-3.5-lite and voyage-3-lite
-    # - 320K for voyage-3.5, voyage-3, and voyage-2
-    # - 120K for voyage-3-large, voyage-code-3, voyage-large-2-instruct, voyage-finance-2, voyage-multilingual-2, voyage-law-2, voyage-large-2, and voyage-3-lite
+    # - 1M for voyage-4-lite, voyage-3.5-lite, and voyage-3-lite
+    # - 320K for voyage-4, voyage-3.5, voyage-3, and voyage-2
+    # - 120K for voyage-4-large, voyage-3-large, voyage-code-3, voyage-large-2-instruct, voyage-finance-2, voyage-multilingual-2, voyage-law-2, voyage-large-2
     match model:
-        case "voyage-3.5-lite" | "voyage-3-lite":
+        case "voyage-4-lite" | "voyage-3.5-lite" | "voyage-3-lite":
             return 1_000_000
-        case "voyage-3.5" | "voyage-2" | "voyage-3":
+        case "voyage-4" | "voyage-3.5" | "voyage-2" | "voyage-3":
             return 320_000
         case _:
             return 120_000  # Conservative default for specialized and older models