fix: fixed summary models for deepseek and openrouter, added TODOs for improving openrouter support

[ErikBjare]

Fixing issues found during #411

---

Important

Fix summary models for deepseek and openrouter, and add TODOs for openrouter improvements.

• Behavior:
  • Fix _summarize_str in __init__.py to correctly handle model context and token limits using get_model().
  • Add TODOs in llm_openai.py for improving provider routing, setting required parameters, and quantization for openrouter.
• Models:
  • Update get_summary_model() in models.py to return deepseek-chat for deepseek provider.

^{This description was created by for 6be1d41. It will automatically update as commits are pushed.}

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 69.97%. Comparing base (07ed85f) to head (6be1d41).

✅ All tests successful. No failed tests found.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #414      +/-   ##
==========================================
- Coverage   69.98%   69.97%   -0.02%     
==========================================
  Files          70       70              
  Lines        5818     5816       -2     
==========================================
- Hits         4072     4070       -2     
  Misses       1746     1746              

Flag	Coverage Δ
anthropic/claude-3-haiku-20240307	68.70% <100.00%> (-0.03%)	⬇️
openai/gpt-4o-mini	68.03% <100.00%> (+0.22%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[ellipsis-dev]

👍 Looks good to me! Reviewed everything up to 6be1d41 in 37 seconds

More details

• Looked at 68 lines of code in 3 files
• Skipped 0 files when reviewing.
• Skipped posting 2 drafted comments based on config settings.

1. gptme/llm/__init__.py:186

• Draft comment:
  Ensure that _chat_complete can handle model.full as input. It expects a model string, and using model.full might lead to unexpected behavior if not handled properly.
• Reason this comment was not posted:
  Comment did not seem useful.

2. gptme/llm/models.py:261

• Draft comment:
  The addition of 'deepseek' in get_summary_model aligns with the changes in the MODELS dictionary, ensuring support for 'deepseek'.
• Reason this comment was not posted:
  Confidence changes required: 0%
  The PR adds a new provider 'deepseek' to the get_summary_model function. This is consistent with the changes made in the MODELS dictionary where 'deepseek' models are defined. The addition of 'deepseek' in get_summary_model ensures that a summary model is returned for this provider, which aligns with the intent of the PR to support 'deepseek'.

Workflow ID: wflow_L1KITvmLdn48cdwO

---

You can customize Ellipsis with 👍 / 👎 feedback, review rules, user-specific overrides, quiet mode, and more.