groq/compound and groq/compound-mini Deprecation
Deprecation of groq/compound and groq/compound-mini. Both systems will be decommissioned on 2026-09-21. No recommended replacement provided.
Groq updates
Last updated · 21 releases tracked since April 2025
A dated log of everything Groq has shipped, collected from official Groq sources and updated as new releases land.
AI infrastructuregroq.com
October 2026 at a glance
No Groq releases yet in October 2026. The last one, “groq/compound and groq/compound-mini Deprecation”, landed on September 21, 2026.
This month
0
-2 vs last month
Last 12 months
12
Busiest: March 2026
Most shipped
Model
9 releases
Latest Groq releases
Newest first. Each entry links to the official page it came from.
September 20262 releases
qwen/qwen3.6-27b Deprecation
Deprecation of qwen/qwen3.6-27b in favor of qwen/qwen3.8-27b. Shutdown on 2026-09-14.
August 20261 release
llama-3.1-8b-instant and llama-3.3-70b-versatile Deprecation
Deprecation of llama-3.1-8b-instant and llama-3.3-70b-versatile. Shutdown on 2026-08-16. Recommended replacements: openai/gpt-oss-20b and openai/gpt-oss-120b or qwen/qwen3.6-27b.
July 20261 release
qwen/qwen3-32b and meta-llama/llama-4-scout-17b-16e-instruct Deprecation
Deprecation of qwen/qwen3-32b and meta-llama/llama-4-scout-17b-16e-instruct. Shutdown on 2026-07-17. Recommended replacements: openai/gpt-oss-120b and openai/gpt-oss-120b or qwen/qwen3.6-27b.
April 20261 release
moonshotai/kimi-k2-instruct-0905 Deprecation
Deprecation of moonshotai/kimi-k2-instruct-0905. Shutdown on 2026-04-15. Recommended replacement: openai/gpt-oss-120b.
March 20264 releases
Model Deprecation: llama-4-maverick-17b-128e-instruct
Groq deprecated llama-4-maverick-17b-128e-instruct in favor of openai/gpt-oss-120b, redirecting users to the newer reasoning model.
ModelSource: Groq Docs
meta-llama/llama-4-maverick-17b-128e-instruct Deprecation
Deprecation of meta-llama/llama-4-maverick-17b-128e-instruct. Shutdown on 2026-03-09. Recommended replacement: openai/gpt-oss-120b.
Model Deprecation: llama-guard-4-12b
Groq deprecated llama-guard-4-12b in favor of openai/gpt-oss-safeguard-20b, keeping trust-and-safety workflows on a newer model.
Enterprise & SecuritySource: Groq Docs
meta-llama/llama-guard-4-12b Deprecation
Deprecation of meta-llama/llama-guard-4-12b. Shutdown on 2026-03-05. Recommended replacement: openai/gpt-oss-safeguard-20b.
February 20261 release
GroqCloud: Expanding to Meet Demand
Groq said GroqCloud demand was accelerating and described new infrastructure, including a UK data center, to keep inference capacity closer to users.
InfrastructureSource: Groq Blog
January 20261 release
Canopy Labs' Orpheus TTS is live on GroqCloud
Groq added Canopy Labs' Orpheus text-to-speech models to GroqCloud, with English and Saudi Arabic variants replacing the older PlayAI endpoints.
December 20251 release
playai-tts and playai-tts-arabic Deprecation
Deprecation of playai-tts and playai-tts-arabic. Shutdown on 2025-12-31. Recommended replacements: canopylabs/orpheus-v1-english and canopylabs/orpheus-arabic-saudi.
Earlier releases (9)
October 20253 releases
moonshotai/kimi-k2-instruct Deprecation
Deprecation of moonshotai/kimi-k2-instruct. Shutdown on 2025-10-10. Recommended replacement: openai/gpt-oss-120b.
gemma2-9b-it Deprecation
Deprecation of gemma2-9b-it. Shutdown on 2025-10-08. Recommended replacement: llama-3.1-8b-instant.
deepseek-r1-distill-llama-70b Deprecation
Deprecation of deepseek-r1-distill-llama-70b. Shutdown on 2025-10-02. Recommended replacements: llama-3.3-70b-versatile or openai/gpt-oss-120b.
August 20252 releases
llama3-70b-8192 and llama3-8b-8192 Deprecation
Deprecation of llama3-70b-8192 and llama3-8b-8192. Shutdown on 2025-08-30. Recommended replacements: llama-3.3-70b-versatile and llama-3.1-8b-instant respectively.
distil-whisper-large-v3-en Deprecation
Deprecation of distil-whisper-large-v3-en. Shutdown on 2025-08-23. Recommended replacement: whisper-large-v3-turbo.
July 20252 releases
mistral-saba-24b Deprecation
Deprecation of mistral-saba-24b. Shutdown on 2025-07-30. Recommended replacement: qwen/qwen3-32b.
qwen-qwq-32b Deprecation
Deprecation of qwen-qwq-32b. Shutdown on 2025-07-14. Recommended replacement: qwen/qwen3-32b.
June 20251 release
llama-guard-3-8b Deprecation
Deprecation of llama-guard-3-8b. Shutdown on 2025-06-06. Recommended replacement: meta-llama/llama-guard-4-12b.
April 20251 release
Multiple Model Deprecations
Deprecation of multiple models including llama-3.2-1b-preview, llama-3.2-3b-preview, and llama-3.2-11b-vision-preview. Shutdown on 2025-04-14. Recommended replacements: llama-3.1-8b-instant and meta-llama/llama-4-scout-17b-16e-instruct.
Shipping cadence
Releases per month over the last 12 months. Busiest month: March 2026, 4 releases.
What Groq ships
Releases in the last 12 months by category.
By category
- Model9
- API1
- Enterprise & Security1
- Infrastructure1
Groq updates FAQ
When did Groq last ship an update?
The most recent Groq update we track is “groq/compound and groq/compound-mini Deprecation”, released on September 21, 2026.
What did Groq release in September 2026?
Groq shipped 2 updates in September 2026, including “groq/compound and groq/compound-mini Deprecation”, “qwen/qwen3.6-27b Deprecation”.
How often does Groq release updates?
We tracked 12 updates from Groq in the last 12 months, about 1 a month. The busiest month was March 2026, with 4 releases.
Related companies
Newsletter
Stay ahead of the month's AI releases without living in every company feed.
This viewscreen turns the calendar into a recurring digest: the sharpest launches, model drops, and product moves delivered in one clear read.
Monthly signal
The standout launches, model drops, and platform moves worth tracking.
Zero feed fatigue
One tight read instead of bouncing between company blogs and timelines.
Built for operators
Clear summaries you can scan fast and forward to the team when needed.
Monthly dispatch
Subscribe to the newsletter
Get a clean monthly dispatch covering the launches that mattered, not every minor product breadcrumb.
Monthly digestMajor launchesFast summaries
One email each month, tuned for signal over noise.