Gemini 3.1 Flash Lite Reaches General Availability
Google released Gemini 3.1 Flash Lite as a GA model on May 7, 2026. This brief details its specs, the deprecation of the preview version, and evaluation steps.
Release Details and Specifications
Google officially released gemini-3.1-flash-lite on May 7, 2026. This event marks the transition from a preview state to general availability (GA). The source describes the model as optimized for speed, scale, and cost efficiency. It is part of the Gemini family and is provided by Google.
The technical specifications listed in the source indicate that the model accepts text and image modalities for input. The output modality is limited to text. The context window size is documented as 1,048,576 tokens, while the maximum output token limit is set at 65,536 tokens. The source records that open weights are not confirmed for this release.
Deprecation of the Preview Model
Concurrent with the GA release, the source notes that the gemini-3.1-flash-lite-preview model has been shut down. Developers are directed to use the new gemini-3.1-flash-lite identifier instead. A specific deprecation announcement states that the preview model was deprecating on May 11, 2026, and would be fully shut down on May 25, 2026.
The changelog confirms that the preview model was launched as the first Flash-Lite model in the Gemini 3 series. However, the current status is defined by the GA release. The source does not provide details on how the performance or behavior of the GA version differs specifically from the preview version beyond the change in availability status.
Evaluation Procedure for Engineers
To evaluate the model, engineers should propose tests based on the documented capabilities. Since the source confirms support for text and image inputs, test cases should include both modalities. Measure the response time and token usage against the stated optimization goals of speed and cost efficiency.
Validate the context window handling by submitting inputs approaching the 1,048,576-token limit. Compare the output quality against existing baselines using standard metrics. Note that the source does not provide benchmark scores or pricing data, so any cost analysis must rely on external documentation not present in this fact set. Do not assume local installation is possible or impossible based solely on the openWeights=false flag; verify access methods through official API documentation.
- Test text and image input processing.
- Measure latency and throughput.
- Validate context window limits.
- Compare output consistency with previous models if available.
Primary sources
Frequently asked questions
What is the exact date of the general availability announcement?
The source states that the general availability release occurred on May 7, 2026. This is distinct from the publication date of this article.
Is the preview model still accessible?
No. The source indicates that the gemini-3.1-flash-lite-preview model has been shut down. Users are instructed to use the gemini-3.1-flash-lite model instead.