Skip to content

Mikec/01 adding gpt 5 #80

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Oct 22, 2025
Merged

Mikec/01 adding gpt 5 #80

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
93765c5
adding support for gpt-5
mikecann Oct 21, 2025
73139db
updated package json to default to gpt5
mikecann Oct 21, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
15 changes: 8 additions & 7 deletions package.json
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -26,13 +26,14 @@
"typescript-eslint": "^8.23.0",
"vitest": "^3.0.2"
},
"scripts": {
"local:run:fundamentals": "cross-env TEST_FILTER=000-fundamentals npm run local:run",
"local:run:one": "cross-env TEST_FILTER=000-fundamentals/000 npm run local:run",
"local:run": "cross-env MODELS=gpt-4.1 DISABLE_BRAINTRUST=1 VERBOSE_INFO_LOGS=1 LOCAL_RESULTS=local_results.jsonl npm run run:evals",
"braintrust:fundamentals": "cross-env TEST_FILTER=000-fundamentals npm run braintrust:run",
"braintrust:run:one": "cross-env TEST_FILTER=000-fundamentals/000 npm run braintrust:run",
"braintrust:run": "cross-env MODELS=gpt-4.1 VERBOSE_INFO_LOGS=1 LOCAL_RESULTS=local_results.jsonl npm run run:evals",
"scripts": {
"local:run:fundamentals": "cross-env MODELS=gpt-5 TEST_FILTER=000-fundamentals npm run local:run",
"local:run:one": "cross-env MODELS=gpt-5 TEST_FILTER=000-fundamentals/000 npm run local:run",
"local:run": "cross-env DISABLE_BRAINTRUST=1 VERBOSE_INFO_LOGS=1 LOCAL_RESULTS=local_results.jsonl npm run run:evals",
"grade:last": "bun run scripts/gradeLast.ts --concise",
"braintrust:fundamentals": "cross-env MODELS=gpt-5 TEST_FILTER=000-fundamentals npm run braintrust:run",
"braintrust:run:one": "cross-env MODELS=gpt-5 TEST_FILTER=000-fundamentals/000 npm run braintrust:run",
"braintrust:run": "cross-env LOCAL_RESULTS=local_results.jsonl VERBOSE_INFO_LOGS=1 npm run run:evals",
"run:evals": "pdm run python -m runner.eval_convex_coding"
}
}
16 changes: 16 additions & 0 deletions runner/models/__init__.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -89,6 +89,22 @@ class ModelTemplate(BaseModel):
uses_system_prompt=False,
provider=ModelProvider.OPENAI,
),
ModelTemplate(
name="gpt-5-mini",
formatted_name="GPT-5 mini",
max_concurrency=int(os.getenv("OPENAI_CONCURRENCY", "4")),
requires_chain_of_thought=False,
uses_system_prompt=False,
provider=ModelProvider.OPENAI,
),
ModelTemplate(
name="gpt-5-nano",
formatted_name="GPT-5 nano",
max_concurrency=int(os.getenv("OPENAI_CONCURRENCY", "4")),
requires_chain_of_thought=False,
uses_system_prompt=False,
provider=ModelProvider.OPENAI,
),
ModelTemplate(
name="deepseek-ai/DeepSeek-V3",
formatted_name="DeepSeek V3",
Expand Down
18 changes: 13 additions & 5 deletions runner/models/model_codegen.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -45,14 +45,22 @@ def generate(self, prompt: str):
system_message = {"role": "system", "content": SYSTEM_PROMPT}
else:
system_message = {"role": "user", "content": SYSTEM_PROMPT}
response = self.client.chat.completions.create(
model=self.model.name,
messages=[
# Build parameters, selecting the correct token limit key for newer models
max_token_limit = 8192 if self.model.name == "claude-3-5-sonnet-latest" else 16384
create_params = {
"model": self.model.name,
"messages": [
system_message,
{"role": "user", "content": user_prompt},
],
max_tokens=8192 if self.model.name == "claude-3-5-sonnet-latest" else 16384,
)
}
# Some newer models (e.g., GPT-5 family) expect `max_completion_tokens` instead of `max_tokens`.
if self.model.name.startswith("gpt-5"):
create_params["max_completion_tokens"] = max_token_limit
else:
create_params["max_tokens"] = max_token_limit

response = self.client.chat.completions.create(**create_params)
return self._parse_response(response.choices[0].message.content)

def _parse_response(self, response: str):
Expand Down