Free LLM API errors · Gemini API · checked 2026-11-13
Gemma: Developer instruction is not enabled
HTTP 400 · Developer instruction is not enabled for models/gemma-3-27b-itYou sent a system instruction (a system message, in OpenAI terms) to a Gemma model whose Gemini API endpoint does not accept one.
Why it happens
- Reported on Google's developer forum for Gemma 3 models such as models/gemma-3-27b-it.
- OpenAI-compatible clients turn the system message into a system instruction, so a plain chat with a system prompt triggers it.
How to fix it
- Move to a Gemma model Google lists for the Gemini API today: gemma-4-31b-it or gemma-4-26b-a4b-it. Google's Gemma page documents system instructions for them.
- Or, for a model that does not accept it, put the instructions at the start of the first user message instead of a system message.
Questions
- Does Gemma support system prompts on the Gemini API?
- Google's Gemma page documents system instructions for gemma-4-31b-it and gemma-4-26b-a4b-it. Older Gemma 3 ids have been reported to reject them.
Sources
Read on 2026-11-13. Providers change limits and messages often; the linked pages are what applies.
Limits and fine print of Gemini API's free tier: Gemini API free API. Tired of handling these by hand? ZeroLimitAI is one OpenAI-compatible endpoint that moves to the next free model when one of these comes back, and can run on your own Groq or OpenRouter key.