Summary
Groq has deprecated many models over the past year. The Groq Deprecation page documents each with explicit dates and replacements, and the Groq Models page now lists only 7 models (4 production + 3 preview). The catalog still includes at least 8 deprecated Groq models. This issue covers the 3 most recent; additional stale entries are listed below for cleanup.
Top 3 Deprecated Models
| Model ID |
Deprecation Date |
Replacement |
Other Providers |
moonshotai/kimi-k2-instruct-0905 |
April 15, 2026 |
openai/gpt-oss-120b |
None (Groq-only) |
meta-llama/llama-4-maverick-17b-128e-instruct |
March 9, 2026 |
openai/gpt-oss-120b |
None (Groq-only) |
gemma2-9b-it |
October 8, 2025 |
llama-3.1-8b-instant |
None (Groq-only) |
All three have Groq as their only provider and should be deprecated or removed from the catalog entirely.
Additional Deprecated Groq Models Still in Catalog
| Model ID |
Deprecation Date |
Other Providers |
deepseek-r1-distill-llama-70b |
October 2, 2025 |
cerebras (also deprecated there — see issue #807) |
llama3-70b-8192 |
August 30, 2025 |
None (Groq-only) |
llama3-8b-8192 |
August 30, 2025 |
None (Groq-only) |
mistral-saba-24b |
July 30, 2025 |
None (Groq-only) |
qwen-qwq-32b |
July 14, 2025 |
None (Groq-only) |
Groq's Current Model Inventory
Per the Groq Models page, only these models are currently available:
Production: llama-3.1-8b-instant, llama-3.3-70b-versatile, openai/gpt-oss-120b, openai/gpt-oss-20b
Preview: meta-llama/llama-4-scout-17b-16e-instruct, openai/gpt-oss-safeguard-20b, qwen/qwen3-32b
Verification Checklist
| Check |
Status |
Detail |
| Cross-source |
✅ |
All models confirmed absent from Groq Models page AND explicitly listed on Groq Deprecation page with past dates |
| Recent commits |
✅ |
No commit in the past 30 days touches Groq provider mappings for these models |
| Duplicate check |
✅ |
No existing open issue covers Groq deprecations |
Local Files Inspected
packages/proxy/schema/model_list.json — confirmed all models still have entriespackages/proxy/schema/index.ts — confirmed AvailableEndpointTypes still maps these models to groq
{
"kind": "deprecated_models",
"provider": "groq",
"models": ["moonshotai/kimi-k2-instruct-0905", "meta-llama/llama-4-maverick-17b-128e-instruct", "gemma2-9b-it"],
"status": "deprecated",
"source_urls": [
"https://console.groq.com/docs/models",
"https://console.groq.com/docs/deprecations"
]
}
Summary
Groq has deprecated many models over the past year. The Groq Deprecation page documents each with explicit dates and replacements, and the Groq Models page now lists only 7 models (4 production + 3 preview). The catalog still includes at least 8 deprecated Groq models. This issue covers the 3 most recent; additional stale entries are listed below for cleanup.
Top 3 Deprecated Models
moonshotai/kimi-k2-instruct-0905openai/gpt-oss-120bmeta-llama/llama-4-maverick-17b-128e-instructopenai/gpt-oss-120bgemma2-9b-itllama-3.1-8b-instantAll three have Groq as their only provider and should be deprecated or removed from the catalog entirely.
Additional Deprecated Groq Models Still in Catalog
deepseek-r1-distill-llama-70bcerebras(also deprecated there — see issue #807)llama3-70b-8192llama3-8b-8192mistral-saba-24bqwen-qwq-32bGroq's Current Model Inventory
Per the Groq Models page, only these models are currently available:
Production:
llama-3.1-8b-instant,llama-3.3-70b-versatile,openai/gpt-oss-120b,openai/gpt-oss-20bPreview:
meta-llama/llama-4-scout-17b-16e-instruct,openai/gpt-oss-safeguard-20b,qwen/qwen3-32bVerification Checklist
Local Files Inspected
packages/proxy/schema/model_list.json— confirmed all models still have entriespackages/proxy/schema/index.ts— confirmedAvailableEndpointTypesstill maps these models togroq{ "kind": "deprecated_models", "provider": "groq", "models": ["moonshotai/kimi-k2-instruct-0905", "meta-llama/llama-4-maverick-17b-128e-instruct", "gemma2-9b-it"], "status": "deprecated", "source_urls": [ "https://console.groq.com/docs/models", "https://console.groq.com/docs/deprecations" ] }