Skip to content

feat: add LLMTR provider#2062

Open
knowhycodata wants to merge 9 commits into
anomalyco:devfrom
knowhycodata:add-llmtr-provider
Open

feat: add LLMTR provider#2062
knowhycodata wants to merge 9 commits into
anomalyco:devfrom
knowhycodata:add-llmtr-provider

Conversation

@knowhycodata

Copy link
Copy Markdown

Adds LLMTR, an OpenAI-compatible AI gateway hosted in Turkey, with its six first-party models (Gemma 4, Qwen 3.6 35B-A3B, MedGemma 4B, Trendyol 7B, Magibu 11B v8, Sincap).

Passthrough models (OpenAI/Anthropic/etc.) are reachable via user config and can be added later.

  • bun validate passes.
  • Pricing/limits/capabilities taken from the per-model pages on llmtr.com.
  • release_date/last_updated derived from the API created timestamps; output limits are conservative defaults where the model page doesn't state one.

Verified with opencode: built an augmented catalog and confirmed it lists all six llmtr/* models and completes a chat with llmtr/gemma-4.

Related opencode issue: anomalyco/opencode#31315

@@ -0,0 +1,5 @@
name = "LLMTR"

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

logo?

Copy link
Copy Markdown
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Added providers/llmtr/logo.svg (adapted from LLMTR's favicon, using currentColor).

Comment thread providers/llmtr/models/qwen3-6-35b.toml Outdated
@@ -0,0 +1,21 @@
name = "Qwen 3.6 35B-A3B"

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

use base_model syntax if possible

Copy link
Copy Markdown
Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Done — qwen3-6-35b now uses base_model = "alibaba/qwen3.6-35b-a3b" and only declares the LLMTR-specific cost + a smaller served context. The other five don't have a clean base: medgemma/trendyol/magibu/sincap aren't in models/ (LLMTR-specific), and gemma-4 has two base variants (gemma-4-31b-it, gemma-4-26b-a4b-it) and LLMTR doesn't say which it serves — happy to map it if you'd prefer one.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants