Skip to content

feat(groq): add Qwen3.8 27B - #5742

Open
DevSuyashA wants to merge 2 commits into
anomalyco:devfrom
DevSuyashA:dev
Open

feat(groq): add Qwen3.8 27B#5742
DevSuyashA wants to merge 2 commits into
anomalyco:devfrom
DevSuyashA:dev

Conversation

@DevSuyashA

@DevSuyashA DevSuyashA commented Aug 28, 2026

Copy link
Copy Markdown

Adds the Qwen3.8 27B model to the Groq provider catalog.

Model data (reported by GET /v1/models):

  • base_model: alibaba/qwen3.8-27b
  • context window: 131,042
  • max output tokens: 16,384
  • cost (USD/MTok): input 0.80, output 4.00, cache_read 0.40
  • modalities (input): text, image
  • reasoning_options: effort [none, default]

Validated with bun validate (import gate passes).

Sources (first-party, accessed 2026-08-29):

Context limit note: context_limit 131,042 confirmed live against GET https://api.groq.com/openai/v1/models for qwen/qwen3.8-27b on 2026-08-29 (returned context_window = 131042). The value is now closed/confirmed — no correction needed.

@github-actions

Copy link
Copy Markdown
Contributor

Action items

  • [medium] [possible mistake] providers/groq/models/qwen/qwen3.8-27b.toml:10 - Check: Provider context limit must match what Groq actually serves. Why: context = 131_042 is unique in the catalog and differs from every other Groq chat model (including the same-surface peer qwen/qwen3.6-27b), which use 131_072. That is a plausible typo or misread of a power-of-two window. Action: Confirm the value against Groq’s model page or live /v1/models for qwen/qwen3.8-27b; if the host is 131,072 tokens, correct it to 131_072.
  • [low] [possible mistake] providers/groq/models/qwen/qwen3.8-27b.toml:4 - Check: Data-changing PRs should cite first-party pricing/docs/API for material facts. Why: Cost (0.80 / 4.00 / cache_read 0.40), limits, and modalities cannot be reviewed from the catalog alone; the PR body only asserts API verification without a mapped Groq pricing or model doc URL. Action: Add a leading top-of-file comment (and/or PR body) with direct Groq sources and what each supports (pricing, context/output, modalities, reasoning_effort values).

@github-actions

Copy link
Copy Markdown
Contributor

Action items

  • [medium] [possible mistake] providers/groq/models/qwen/qwen3.8-27b.toml:10 - Check: Provider limit.context should match this host’s published/API context window. Why: context = 131_042 is not a power-of-two window and differs from the closest Groq peer (qwen/qwen3.6-27b uses 131_072) and every other standard Groq text model at 128k (131_072). That pattern strongly suggests a typo (e.g. 0 vs 7) rather than an intentional host limit. Action: Confirm against Groq’s model docs or /v1/models and set context to the real value (likely 131_072 unless Groq truly advertises 131_042).

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.

1 participant