KeyError: 'max_tokens' when running ontogpt list-models

[jinzhang6688]

When running ontogpt list-models, the CLI crashes with a KeyError: 'max_tokens' for certain providers (e.g., Amazon Bedrock, TwelveLabs embeddings). These models don’t expose a max_tokens field, but the CLI assumes it exists.

---

Steps to Reproduce

1. Install OntoGPT (latest release).
2. Run:

   ontogpt list-models

3. Observe output: models are listed, then the command fails with:

   KeyError: 'max_tokens'
   

---

Traceback (excerpt)

File "ontogpt/cli.py", line 2228, in list_models
    max_tokens = models[model]["max_tokens"]
KeyError: 'max_tokens'

---

Expected Behavior
ontogpt list-models should list all available models without crashing, even if some providers don’t include max_tokens.

---

Current Code (before fix)

for model in models:
    primary_name = model
    provider = models[model]["litellm_provider"]

    if "mode" in models[model]:
        functionality = models[model]["mode"]
        if functionality not in ["chat", "completion", "embedding"]:
            continue
    else:
        continue

    max_tokens = models[model]["max_tokens"]

---

Proposed Fix
Use a safe lookup for max_tokens:

for model in models:
    model_info = models[model]
    primary_name = model
    provider = model_info.get("litellm_provider", "N/A")

    functionality = model_info.get("mode")
    if functionality not in ["chat", "completion", "embedding"]:
        continue

    max_tokens = model_info.get("max_tokens", "N/A")

---

Environment

• Python 3.13
• OntoGPT (latest pip release)
• litellm (latest pip release)

---

Additional Notes
After applying the .get("max_tokens") patch locally, the command works correctly and lists all models. Suggest updating the CLI to handle missing fields gracefully.