vet

commit 4a6023e01e24b45a9cddc97e28800ffeb7b5222a
parent 71ddc8431d26a07d14c0dc2e4e84c52c74154e75
Author: Andrew Laack <andrew@laack.co>
Date:   Thu,  5 Mar 2026 20:08:23 +0000

Merge pull request #174 from imbue-ai/andrew/gpt-5-4

GPT 5.4 Support
Diffstat:
M
registry/models.json
 | 
18
++++++++++++++++++
M
uv.lock
 | 
2
+-
M
vet/imbue_core/agents/llm_apis/openai_api.py
 | 
20
++++++++++++++++++++

3 files changed, 39 insertions(+), 1 deletion(-)
diff --git a/registry/models.json b/registry/models.json
@@ -1,5 +1,23 @@
 {
   "providers": {
+    "openai": {
+      "name": "OpenAI",
+      "api_type": "openai_compatible",
+      "base_url": "https://api.openai.com/v1",
+      "api_key_env": "OPENAI_API_KEY",
+      "models": {
+        "gpt-5.4": {
+          "context_window": 1050000,
+          "max_output_tokens": 128000,
+          "supports_temperature": false
+        },
+        "gpt-5.4-pro": {
+          "context_window": 1050000,
+          "max_output_tokens": 128000,
+          "supports_temperature": false
+        }
+      }
+    },
     "groq": {
       "name": "Groq",
       "api_type": "openai_compatible",
diff --git a/uv.lock b/uv.lock
@@ -1494,7 +1494,7 @@ wheels = [
 
 [[package]]
 name = "verify-everything"
-version = "0.2.3"
+version = "0.2.4"
 source = { editable = "." }
 dependencies = [
     { name = "anthropic" },
diff --git a/vet/imbue_core/agents/llm_apis/openai_api.py b/vet/imbue_core/agents/llm_apis/openai_api.py
@@ -63,6 +63,8 @@ class OpenAIModelName(enum.StrEnum):
     GPT_5_MINI = "gpt-5-mini"
     GPT_5_1 = "gpt-5.1"
     GPT_5_2 = "gpt-5.2"
+    GPT_5_4 = "gpt-5.4"
+    GPT_5_4_PRO = "gpt-5.4-pro"
 
 
 # Using Tier 5 rate limits
@@ -142,6 +144,22 @@ OPENAI_MODEL_INFO_BY_NAME: FrozenMapping[OpenAIModelName, ModelInfo] = FrozenDic
             max_output_tokens=128_000,
             rate_limit_req=15000 / 60,  # 15000 RPM = 250 RPS
         ),
+        OpenAIModelName.GPT_5_4: ModelInfo(
+            model_name=str(OpenAIModelName.GPT_5_4),
+            cost_per_input_token=2.50 / 1_000_000,
+            cost_per_output_token=15 / 1_000_000,
+            max_input_tokens=1_050_000,
+            max_output_tokens=128_000,
+            rate_limit_req=15000 / 60,  # 15000 RPM = 250 RPS
+        ),
+        OpenAIModelName.GPT_5_4_PRO: ModelInfo(
+            model_name=str(OpenAIModelName.GPT_5_4_PRO),
+            cost_per_input_token=30 / 1_000_000,
+            cost_per_output_token=180 / 1_000_000,
+            max_input_tokens=1_050_000,
+            max_output_tokens=128_000,
+            rate_limit_req=10000 / 60,  # 10000 RPM = 166.67 RPS
+        ),
     }
 )
 
@@ -191,6 +209,8 @@ def is_openai_reasoning_model(model_name: str) -> bool:
         OpenAIModelName.GPT_5_MINI,
         OpenAIModelName.GPT_5_1,
         OpenAIModelName.GPT_5_2,
+        OpenAIModelName.GPT_5_4,
+        OpenAIModelName.GPT_5_4_PRO,
     )