Jadual kad model DeepMind. · sumber
Jadual kad model DeepMind. · sumber
Jadual kad model DeepMind. · sumber
Jadual kad model DeepMind. · sumber
| Model | Input | Keluaran |
|---|---|---|
| Gemini 3.5 Flash-Lite | $0.30 | $2.50 |
| Gemini 3.1 Flash-Lite | $0.25 | $1.50 |
| GPT-5.4 mini | $0.75 | $4.50 |
| Claude Haiku 4.5 | $1.00 | $5.00 |
Seperti yang disenaraikan dalam kad model DeepMind. · sumber
Unofficial, from StatusGator. Uptime is the share of the window not marked down (529 min down, 5676 min degraded); it covers the whole service, not this version alone.
Estimate from the version's published specs, benchmarks and reviews, 0-5; scores are only comparable within one criteria year. Scored as launched (Gemini 3.5 Flash-Lite, 21 Jul 2026): Artificial Analysis Intelligence Index 36 (v4.1, high reasoning; up 11 from 3.1 Flash-Lite's 25; AA's current page shows 22 on v4.3.2, not used); 1M-token context; $0.30 in / $2.50 out per 1M = $0.85 blended at 3:1; 350 output tokens/s measured by Artificial Analysis pre-launch (AA's live page later showed about 430-490 t/s, not used); text, image, video, speech input. Coding null: no Terminal-Bench 4.0 at launch (AA measured 53.6% on Terminal-Bench v2.1; Google 54.0%), different benchmark version. Reviews: AA's note, eesel and Vorp Labs commentary (vendor/reseller blogs, not hands-on tests), 9to5Google; no genuine hands-on test found. blog.google
Gemini 3.5 Flash-Lite scores 36 on the Artificial Analysis Intelligence Index, up 11 points from Gemini 3.1 Flash-Lite (25)
It is not the smartest model Google ships, and it is not trying to be.
Pilot 3.5 Flash-Lite where throughput, latency, or unit cost matters more than frontier quality
It's a significant step up in coding and agentic tasks as seen in Terminal-Bench 2.1 (54% vs 31%)
| Gemini 3.6 Flash and Gemini 3.5 Flash-Lite: Halving Time per Task · 21 Jul 2026 | 0.70 |
|---|---|
| Gemini 3.5 Flash-Lite model release review · 21 Jul 2026 | 0.40 |
| Google launches Gemini 3.6 Flash and 3.5 Flash-Lite, teases Gemini 4 · 21 Jul 2026 | 0.60 |
| Gemini 3.5 Flash-Lite: what it is, price, and who it's for · 6 Ogo 2026 | 0.20 |
Foto daripada pembaca, berasingan daripada media bersumber pin. Nilaikan apa yang anda makan atau lihat, dan pada pin restoran, tag hidangan terus pada foto anda.
Entri pertama sentiasa sumber pin. Keyakinan keseluruhan ialah purata berpemberat bagi sejauh mana setiap rujukan menyokong masa mula dan tamat yang digunakan di atas; sesebuah rujukan diambil kira separuh kurang bagi setiap 180 hari lebih lama daripada yang terbaharu.
Laporan pelancaran 9to5Google[2] bertarikh 21 Julai 2026 dan menyatakan Google "turut mengumumkan Gemini 3.5 Flash-Lite"; kad model 3.5 Flash-Lite DeepMind diterbitkan pada 21 Julai 2026.
9to5Google's 21 July 2026 launch report gives the prices and the comparison with 3.1 Flash-Lite and Gemini 3 Flash.
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]
Ada yang tertinggal atau salah? Nyatakan dengan kata-kata anda sendiri: pautan yang menyokong pin ini, tarikh mula atau tamat yang berbeza beserta sebabnya, atau fakta yang tiada atau tersilap. AI menyemaknya berbanding sumber pin ini, mencari yang lebih baik, dan menambah mana-mana halaman yang menyokong anda. Sumber pin itu sendiri tetap paling diambil kira. Gambar yang menunjukkan sesuatu yang lain, atau menunjukkannya dengan buruk, turut dilihat, dan dialihkan ke bawah atau diganti.
Tajuk asal: Gemini 3.5 Flash-Lite Released
| Penanda aras | Gemini 3.5 Flash-Lite | Gemini 3.1 Flash-Lite | GPT-5.4 mini | Claude Haiku 4.5 |
|---|---|---|---|---|
| Harga input ($/1M token, tanpa caching) | $0.30 | $0.25 | $0.75 | $1.00 |
| Harga output ($/1M token) | $2.50 | $1.50 | $4.50 | $5.00 |
| SWE-Bench Pro (Public) (Pelbagai tugas pengekodan agentik) | 54.2% | 38.3% | 54.4% | 39.5% |
| Terminal-bench 2.1 (Pengekodan terminal agentik, harness Terminus-2) | 54.0% | 31.0% | 59.2% | 44.2% |
| MLE-Bench (Kejuruteraan Pembelajaran Mesin) | 39.2% | 22.0% | — | — |
| GDPVal-AA v2 (Kerja pengetahuan, Elo) | 1140 | 642 | 1171 | 907 |
| OSWorld-Verified (Penggunaan komputer agentik) | 74.0% | 54.3% | 72.1% | 50.7% |
| CharXiv Reasoning (Sintesis maklumat daripada carta kompleks, tanpa alat) | 74.5% | 73.2% | 80.3% | 61.7% |
| CharXiv Reasoning (Sintesis maklumat daripada carta kompleks, dengan alat) | 76.5% | 75.6% | — | — |
| GDM-MRCR v2 (8-needle) (Prestasi konteks panjang, 128k (purata)) | 72.2% | 60.1% | 42.7% | 35.3% |
| GDM-MRCR v2 (8-needle) (Prestasi konteks panjang, 1M (titik demi titik)) | 21.3% | 12.3% | — | — |