ttyS3 fc09346ae7 chore(vultr): sync model catalog to latest API ...

- Update Kimi-K2.6 pricing to 0.30/1.20 (was 0.15/0.60)
- Replace GLM-5.1-FP8 with GLM-5.2-FP8 (base_model zhipuai/glm-5.2,
  limit.context 393_216 per vLLM max_model_len)
- Add Qwen3.5-397B-A17B (0.30/2.00)
- Add Qwen3.6-27B (0.30/2.00)
- Add MiMo-V2.5-Pro (0.55/1.65)
- Add DeepSeek-V4-Flash (0.30/1.00)
- Remove Llama-3.1-Nemotron-Safety-Guard-8B-v3 (no longer in API)

All reasoning models declare reasoning_options = [] per Vultr's
lack of reasoning request control. New TextGeneration models use
base_model inheritance from models/ metadata. ReRank models
(VultronRetriever*) are intentionally skipped pending official
documentation of their capabilities.

Source: https://api.vultrinference.com/v1/models
Co-authored-by: Cursor <cursoragent@cursor.com>

2026-07-07 21:27:28 +08:00

..

models

chore(vultr): sync model catalog to latest API

2026-07-07 21:27:28 +08:00

logo.svg

fix: visually align provider logos

2025-12-21 22:08:18 +01:00

provider.toml

docs: document provider reasoning request formats

2026-06-26 09:37:21 -05:00