Gemini 3.5 Flash-Lite

2026-07-21 · Publisher : Google · Domain : Artificial Intelligence

Google launches Gemini 3.5 Flash-Lite, its fastest and most cost-effective model in the 3.5 series, at about 350 tokens per second, for agentic search, document processing and high-throughput workloads.

### Specifications - Context: up to 1 million tokens; output up to 64,000 tokens; text, image, audio and video inputs. - Knowledge cutoff: March 2026. - Based on Gemini 3.1 Flash-Lite. - API identifier: gemini-3.5-flash-lite (changelog of July 21, 2026). ### Pricing | Item | Standard (USD / 1M tokens) | Batch | |---|---|---| | Input | 0.30 | 0.15 | | Output | 2.50 | 1.25 | ### Benchmarks | Benchmark (model card, scores published by the vendor) | 3.5 Flash-Lite | 3.1 Flash-Lite | GPT-5.4 mini | Claude Haiku 4.5 | |---|---|---|---|---| | SWE-Bench Pro | 54.2% | 38.3% | 54.4% | 39.5% | | Terminal-Bench 2.1 | 54.0% | 31.0% | 59.2% | 44.2% | | OSWorld-Verified | 74.0% | 54.3% | 72.1% | 50.7% | | GDPval-AA v2 (Elo) | 1140 | 642 | 1171 | 907 | | CharXiv Reasoning | 74.5% | 73.2% | 80.3% | 61.7% | ### Key takeaways The entry-level tier for high-volume agents, with agent scores close to those of much more expensive models.

Tags : LLM, Google, Gemini, Flash-Lite, Faible coût