Groq
llama-3.3-70b-versatile model retirement
Stopped workingSubscribe to GroqShare
Calls to model llama-3.3-70b-versatile fail on free and developer tiers since August 16, 2026. Groq shut down access to the model on that day for these tiers, causing requests using it to fail rather than redirect to a newer model.
The error you get
What Groq answers a call to llama-3.3-70b-versatile once the model is gone, as recorded from real responses:
{"error":{"message":"The model `llama-3.3-70b-versatile` does not exist or you do not have access to it.","type":"invalid_request_error","code":"model_not_found"}}What to use instead of llama-3.3-70b-versatile
Move to openai/gpt-oss-120b. The other recommended replacement, qwen/qwen3.6-27b, stopped working on September 14, 2026.
In Groq's words
Deprecated Model: llama-3.3-70b-versatile; Shutdown Date: 08/16/26; Recommended Replacement Model ID: openai/gpt-oss-120b or qwen/qwen3.6-27b
Facts
- API
- Groq
api.groq.com - Model
llama-3.3-70b-versatile- Released
- December 6, 2024 (models.dev)
- Kind
- Model retirement
- Takes effect
- August 16, 2026
- Calls after that day
- Fail
- Who it affects
- free and developer tiers
- Source
- Deprecations page
- Last updated
- Verified
- , against Groq's own pages
Questions
When does llama-3.3-70b-versatile shut down?
Calls to the model llama-3.3-70b-versatile on Groq stopped working on August 16, 2026.
What happens to calls after August 16, 2026?
They have failed since August 16, 2026.
What error does llama-3.3-70b-versatile return once it is shut down?
Groq answers HTTP 404: "The model `llama-3.3-70b-versatile` does not exist or you do not have access to it.". Send another model to fix it.
When was llama-3.3-70b-versatile released?
llama-3.3-70b-versatile was released on December 6, 2024, according to models.dev.
How do I find the calls this hits in my code?
Okapi reads your repository, finds the calls this change hits and opens a pull request with the fix. It never commits to your branches: every change comes as a pull request you review.
More from Groq 10
-
playai-ttsModel retirement Stopped working -
meta-llama/llama-guard-4-12bModel retirement Stopped working -
meta-llama/llama-4-maverick-17b-128e-instructModel retirement Stopped working -
moonshotai/kimi-k2-instruct-0905Model retirement Stopped working -
meta-llama/llama-4-scout-17b-16e-instructModel retirement Stopped working -
qwen/qwen3-32bModel retirement Stopped working -
llama-3.1-8b-instantModel retirement Stopped working -
qwen/qwen3.6-27bModel retirement Stopped working -
groq/compoundModel retirement Stopped working -
groq/compound-miniModel retirement Stopped working
You fixed this one by hand.
Okapi opens the pull request with the fix before the next breaking change reaches your code.