- Start
- Jul 21, 202690% CONFIDENCEfrom the source
Gemini 3.5 Flash-Lite Released
- Google announced Gemini 3.5 Flash-Lite on 21 July 2026 for high-throughput, low-latency tasks such as agentic search and document processing, with "significantly better quality than 3.1 Flash-Lite" from March[2]
- It outperforms Gemini 3 Flash on SWE-Bench Pro (54.2% vs 49.6%) and OSWorld-Verified (74.0% vs 65.1%), and is also coming to Search[2]
Notable features
- Priced at $0.30 per million input tokens and $2.50 per million output tokens, against $0.25 and $1.50 for 3.1 Flash-Lite[1][2]
- Terminal-bench 2.1 54.0% (31.0% for 3.1 Flash-Lite), GDM-MRCR v2 at 128k 72.2% (60.1%) and GDPval-AA v2 1140 Elo (642)[1][2]
- Available through Google Antigravity, AI Studio and Android Studio[2]
Benchmarks[1]
| Benchmark | Gemini 3.5 Flash-Lite | Gemini 3.1 Flash-Lite | GPT-5.4 mini | Claude Haiku 4.5 |
|---|---|---|---|---|
| Input price ($/1M tokens, no caching) | $0.30 | $0.25 | $0.75 | $1.00 |
| Output price ($/1M tokens) | $2.50 | $1.50 | $4.50 | $5.00 |
| SWE-Bench Pro (Public) (Diverse agentic coding tasks) | 54.2% | 38.3% | 54.4% | 39.5% |
| Terminal-bench 2.1 (Agentic terminal coding, Terminus-2 harness) | 54.0% | 31.0% | 59.2% | 44.2% |
| MLE-Bench (Machine Learning Engineering) | 39.2% | 22.0% | — | — |
| GDPVal-AA v2 (Knowledge work, Elo) | 1140 | 642 | 1171 | 907 |
| OSWorld-Verified (Agentic computer use) | 74.0% | 54.3% | 72.1% | 50.7% |
| CharXiv Reasoning (Information synthesis from complex charts, No tools) | 74.5% | 73.2% | 80.3% | 61.7% |
| CharXiv Reasoning (Information synthesis from complex charts, With tools) | 76.5% | 75.6% | — | — |
| GDM-MRCR v2 (8-needle) (Long context performance, 128k (average)) | 72.2% | 60.1% | 42.7% | 35.3% |
| GDM-MRCR v2 (8-needle) (Long context performance, 1M (pointwise)) | 21.3% | 12.3% | — | — |
References 382% CONFIDENCEOverall confidence: 82%How well the pin's source and references back up its dates.Weighted average of how firmly 3 references, the source included, support the pin's start and end times; a reference counts half as much for every 180 days older than the newestShow all pins at 75% confidence or better
The first entry is always the pin's source. Overall confidence is a weighted average of how firmly each reference supports the start and end times used above; a reference counts half as much for every 180 days older than the newest.
- [1]90%deepmind.google/models/model-cards/gemini-3-5-flash-litedeepmind.google· Posted Sep 30, 2026· Starts Jul 21, 2026 ✓· 40% of score
9to5Google's[2] launch report is dated 21 July 2026 and says Google "also announced Gemini 3.5 Flash-Lite"; DeepMind's 3.5 Flash-Lite model card is published 21 July 2026.
- [2]85%Google launches Gemini 3.6 Flash and 3.5 Flash-Lite, teases Gemini 49to5google.com· Published Jul 21, 2026· Starts Jul 21, 2026· 30% of score
9to5Google's 21 July 2026 launch report gives the prices and the comparison with 3.1 Flash-Lite and Gemini 3 Flash.
- [3]70%Neural Newscast: Google Ships Three New Gemini Flash Models [Model Behavior]podcasts.apple.com· Published Jul 21, 2026· 30% of score
Neural Newscast's episode "Google Ships Three New Gemini Flash Models [Model Behavior]" (2026-07-21) says: "with enterprise software. [00:48] Thatcher Collins: I'm Thatcher Collins. [00:49] Thatcher Collins: Nina, the timing of these releases is certainly noteworthy. [00:53] Thatcher Collins: Today , Google shipped Gemini 3.6 Flash, 3.5 Flash Lite, and a specialized security-focused version called Flash Cyber. [01:03] Thatcher Collins: This blitz comes immediately after"[1]
Suggest a correction
Something missing or wrong? Say it in your own words: a link that backs this pin up, a different start or end date and why, or a fact it lacks or gets wrong. The AI checks it against this pin's sources, searches for better ones, and adds any page that backs you up. The pin's own sources still count most. A picture that shows something else, or shows it badly, is looked at too, and moved down or replaced.
