Johannes du Plessis
|
a26d5d32cb
|
feat: Add Fireworks model provider (#1415)
* feat: add Fireworks model provider with accurate reasoning levels
Adds Kimi K2.6, DeepSeek V4 Pro, Nemotron 3 Ultra, GLM 5.1 via fireworks: provider, mapping per-model reasoning_effort. Surfaces each model's real effort range (incl. openai none, deepseek xhigh/max).
* fix: drop serverless-unsupported Nemotron 3 Ultra, document FIREWORKS_API_KEY
---------
Co-authored-by: open-swe[bot] <215916821+open-swe[bot]@users.noreply.github.com>
|
2026-06-04 21:59:47 +00:00 |
|