# CWEval: gpt-5-mini

docker pull co1lin/cweval
docker run --name cweval --rm -it --net host \
  -e OPENAI_KEY \
  -v /Users/tueldt1.re/Documents/safecode/CWEval/evals:/home/ubuntu/CWEval/evals \
  co1lin/cweval zsh

Inside Docker:

cd /home/ubuntu/CWEval
source .env
pip install tenacity

export OPENAI_API_KEY="$OPENAI_KEY"
export OPENAI_API_BASE="https://gateway.ai.cloudflare.com/v1/def31e2cf1530789c604bdaa2abbfcf1/openai-proxy/compat"

Full run:

```bash
python cweval/generate.py gen \
  --model openai/openai/gpt-5-mini \
  --api_base "$OPENAI_API_BASE" \
  --eval_path evals/eval_gpt5mini_t1_n3 \
  --n 3 --num_proc 4 --temperature 1

python cweval/evaluate.py pipeline \
  --eval_path evals/eval_gpt5mini_t1_n3 \
  --num_proc 20 --docker False \
  2>&1 | tee evals/eval_gpt5mini_t1_n3/pipeline.log

python cweval/evaluate.py pipeline \
  --eval_path evals/eval_gpt5mini_t1_n3 \
  --num_proc 20 --docker False

python cweval/evaluate.py report_pass_at_k \
  --eval_path evals/eval_gpt5mini_t1_n3
```



docker run --name cweval --rm -it --net host \
  -e OPENAI_KEY \
  -v /Users/tueldt1.re/Documents/safecode/CWEval:/home/ubuntu/CWEval \
  co1lin/cweval zsh

source .env
pip install tenacity
export OPENAI_API_KEY="$OPENAI_KEY"
export OPENAI_API_BASE="https://gateway.ai.cloudflare.com/v1/def31e2cf1530789c604bdaa2abbfcf1/openai-proxy/compat"

python cweval/evaluate.py pipeline \
  --eval_path evals/eval_gpt5mini_t1_n3 \
  --num_proc 20 --docker False