Kad model Google DeepMind, keputusan pada Februari 2026. · sumber
Percubaan tunggal; kad tidak memberikan angka Claude untuk penanda aras ini. · sumber
| item | Nilai |
|---|---|
| Input | Teks, imej, audio, video; tetingkap konteks sehingga 1M token |
| Keluaran | Teks, 64K token |
| berdasarkan | Gemini 3 Pro |
| Peperiksaan Terakhir Kemanusiaan (tiada alat) | 44.4% (Gemini 3 Pro 37.5%) |
| GPQA Diamond (tiada alatan) | 94.3% |
| SWE-Bench Disahkan (percubaan tunggal) | 80.6% |
| Terminal-Bench 2.0 (Terminus-2) | 68.5% |
| tau2-bench Telecom | 99.3% |
| BrowseComp | 85.9% |
| Atlas MCP | 69.2% |
| MRCR v2 8-jarum, 1M mengikut arah | 26.3% |
Angka Google sendiri, keputusan setakat Februari 2026. Kad itu tidak menyenaraikan harga API. · sumber
Unofficial, from StatusGator. Uptime is the share of the window not marked down (529 min down, 5536 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 (19 Feb 2026, preview). Reasoning: Artificial Analysis Intelligence Index 57 at launch (v4.0; four points ahead of Claude Opus 4.6; The Decoder 21 Feb 2026). AA's pages now show lower numbers (30, 46) after later index versions, not used. Coding null: no Terminal-Bench 4.0; Google published Terminal-Bench 2.0 68.5% (Terminus-2) and SWE-Bench Verified 80.6%. Context 1M input, 64k output (DeepMind model card). Price $2 in / $12 out per 1M up to 200k = $4.50 blended (AA, Google). Speed null: no launch-time speed found (AA now lists 132 tokens/s, undated). Input text, image, audio, video.[1]
The model ranks first in six of ten categories, including agent-based coding, knowledge, scientific reasoning, and physics.
While the leap in logical reasoning and coding benchmarks is impressive, the 'soul' of the model — its emotional depth, empathy, creative flexibility, and nuance — seems to have been significantly reduced.
It is not necessarily faster, and in some modes, it is intentionally slower, taking a moment to chew on a problem before spitting out an answer.
Winner: Gemini wins for providing phrasing I'd actually use because it feels the most natural and immediately usable.
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.
Hantaran Google "Gemini 3.1 Pro: A smarter model for your most complex tasks" bertarikh 19 Februari 2026 dan menyatakan ia dilancarkan pada hari itu kepada pembangun, perusahaan dan pengguna.
The Wikipedia article lists Gemini 3.1 Pro as released in preview on 19 February 2026.
DeepMind's model card gives the benchmark table against the model's predecessor and named rivals.
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 3.1 Pro Released
| Penanda aras | Gemini 3.1 Pro | Gemini 3 Pro | Sonnet 4.6 | Opus 4.6 | GPT-5.2 | GPT-5.3-Codex |
|---|---|---|---|---|---|---|
| Humanity's Last Exam (Penaakulan akademik (set penuh, teks + MM), tanpa alat) | 44.4% | 37.5% | 33.2% | 40.0% | 34.5% | — |
| Humanity's Last Exam (Penaakulan akademik (set penuh, teks + MM), Carian (senarai sekat) + Kod) | 51.4% | 45.8% | 49.0% | 53.1% | 45.5% | — |
| ARC-AGI-2 (Teka-teki penaakulan abstrak, ARC Prize Verified) | 77.1% | 31.1% | 58.3% | 68.8% | 52.9% | — |
| GPQA Diamond (Pengetahuan saintifik, tanpa alat) | 94.3% | 91.9% | 89.9% | 91.3% | 92.4% | — |
| Terminal-Bench 2.0 (Pengekodan terminal agentik, harness Terminus-2) | 68.5% | 56.9% | 59.1% | 65.4% | 54.0% | 64.7% |
| Terminal-Bench 2.0 (Pengekodan terminal agentik, harness lain terbaik yang dilaporkan sendiri) | — | — | — | — | 62.2% | 77.3% |
| SWE-Bench Verified (Pengekodan agentik, percubaan tunggal) | 80.6% | 76.2% | 79.6% | 80.8% | 80.0% | — |
| SWE-Bench Pro (Public) (Pelbagai tugas pengekodan agentik, percubaan tunggal) | 54.2% | 43.3% | — | — | 55.6% | 56.8% |
| LiveCodeBench Pro (Masalah pengekodan pertandingan daripada Codeforces, ICPC dan IOI, Elo) | 2887 | 2439 | — | — | 2393 | — |
| SciCode (Pengekodan penyelidikan saintifik) | 59% | 56% | 47% | 52% | 52% | — |
| APEX-Agents (Tugas profesional jangka panjang) | 33.5% | 18.4% | — | 29.8% | 23.0% | — |
| GDPval-AA Elo (Tugas pakar) | 1317 | 1195 | 1633 | 1606 | 1462 | — |
| τ2-bench (Agentik dan penggunaan alat, Runcit) | 90.8% | 85.3% | 91.7% | 91.9% | 82.0% | — |
| τ2-bench (Agentik dan penggunaan alat, Telekomunikasi) | 99.3% | 98.0% | 97.9% | 99.3% | 98.7% | — |
| MCP Atlas (Aliran kerja berbilang langkah menggunakan MCP) | 69.2% | 54.1% | 61.3% | 59.5% | 60.6% | — |
| BrowseComp (Carian agentik, Carian + Python + Layar) | 85.9% | 59.2% | 74.7% | 84.0% | 65.8% | — |
| MMMU-Pro (Pemahaman dan penaakulan multimodal, tanpa alat) | 80.5% | 81.0% | 74.5% | 73.9% | 79.5% | — |
| MMMLU (Soal jawab berbilang bahasa) | 92.6% | 91.8% | 89.3% | 91.1% | 89.6% | — |
| MRCR v2 (8-needle) (Prestasi konteks panjang, 128k (purata)) | 84.9% | 77.0% | 84.9% | 84.0% | 83.8% | — |
| MRCR v2 (8-needle) (Prestasi konteks panjang, 1M (titik demi titik)) | 26.3% | 26.3% | — | — | — | — |