Google Launches Gemini 3.1 Flash Lite at $0.25/M Input: Half-Price Flash With Thinking Levels and 1M Context
Google launched Gemini 3.1 Flash Lite to general availability on May 7, 2026, at $0.25 per million input tokens and $1.50 per million output tokens. The model is positioned as the cost floor of the Gemini 3 family — priced at half the rate of Gemini 3 Flash while adding a capability that none of its Gemini 2 predecessors had: configurable thinking levels.
Pricing
| Model | Input ($/M) | Output ($/M) |
|---|---|---|
| Gemini 3.1 Flash Lite | $0.25 | $1.50 |
| Gemini 3 Flash | ~$0.50 | ~$3.00 |
| Gemini 2.5 Flash Lite | (prev gen) | (prev gen) |
At weighted-average pricing across Google AI Studio providers, blended input lands at approximately $0.155/M due to cache hit rates in production. Output stays at $1.50/M flat.
Cache tokens are priced at $0.025/M input. Audio cache runs $0.05/M. Per-request overhead: $0.014.
Thinking Levels
The addition of thinking levels to the Flash Lite tier is the structural change. Gemini 3.1 Flash Lite supports four configurations: minimal, low, medium, and high. This mirrors the controls available in Gemini 3 Flash and Pro, and allows developers to dial the cost-quality tradeoff on a per-request basis rather than choosing a different model.
A workflow that uses minimal thinking for classification and high thinking for complex reasoning can stay on a single model at a single rate card, rather than routing between model tiers. That simplification has real infrastructure value at scale.
Capabilities
The model is fully multimodal: text, image, video, audio, and PDF inputs are all supported. Context window is 1,048,576 tokens — the full 1M that Google has standardised across the Gemini 3 family.
Google describes it as optimised for high-volume, low-latency workloads: “lightweight agentic workflows, simple data extraction, and applications where responsiveness and API cost are the primary constraints.”
Performance relative to its generation: Google says it outperforms Gemini 2.5 Flash Lite on overall quality and approaches Gemini 2.5 Flash performance. Specific benchmark numbers were not published at launch. Improvements are documented in audio input and ASR, RAG snippet ranking, translation, data extraction, and code completion.
Developer Economics
For high-throughput agentic pipelines where most tasks don’t require deep reasoning, Gemini 3.1 Flash Lite at $0.25/M input competes directly with the cheapest frontier-tier models. DeepSeek V4 Flash was previously the cost benchmark for this class at $0.14/M; Gemini 3.1 Flash Lite is more expensive but brings Google’s infrastructure reliability, 1M context, and the thinking levels dial.
For teams running Gemini 3 Flash at volume, the arithmetic is simple: the same task on Flash Lite with minimal thinking costs roughly half as much. The question is whether quality holds — and Google has not published the numbers that would answer that definitively.
Key Numbers
- Launch date: May 7, 2026
- Input price: $0.25/M tokens
- Output price: $1.50/M tokens
- Cache hit price: $0.025/M input tokens
- Context window: 1,048,576 tokens
- Thinking levels: minimal, low, medium, high
- Modalities: text, image, video, audio, PDF