Google DeepMindproprietary
Released: 2026-02-19Artificial AnalysisGemini 3.1 Pro
Gemini 3.1 Pro Preview is Google’s frontier reasoning model, delivering enhanced software engineering performance, improved agentic reliability, and more efficient token usage across complex workflows. Building on the multimodal foundation...
Intelligence Index
—/ 100
Artificial Analysis Intelligence Index (independent composite)Throughput Speed
112tok/sec
Output streaming throughputTime to First Answer
16s
Time to first answer token, including any thinkingCost to Run the Index
—USD
Artificial Analysis cost of the full Index runMeasured results · Artificial Analysis
Independent evaluations of Gemini 3.1 Pro. Values are reported as published; nothing here is estimated or derived. Snapshot 2026-11-15.
- Intelligence Index
- —
- Cost to run the Index
- —
- Time per task
- —
- Omniscience index
- —
- Reasoning model
- Yes
- Cached-input price
- $0.2 / 1M
Reasoning & knowledge
- CritPt17.7%
- GPQA Diamond94.1%
- HLE-Diamond31.4%
- Humanity's Last Exam47.0%
- IFBench77.1%
Coding & terminal agents
- SciCode58.7%
- Terminal-Bench 2.173.8%
- Terminal-Bench 4.04.0%
- Terminal-Bench Hard53.8%
Agentic work
- AA-AnalystAgent41.3%
- AA-Briefcase v1.1 (Elo)457 Elo
- APEX-Agents-AA32.0%
- AutomationBench-AA35.4%
- EnterpriseOps-Gym-AA42.2%
- GDPval-AA (normalized Elo)799 Elo
- Harvey LAB-AA0.0%
- ITBench-AA30.3%
- τ²-Bench Telecom95.6%
- τ³-Banking21.4%
Long context & documents
Multimodal & science
- MMMU-Pro82.4%
Headline Evaluation Breakdown
Six independently measured evaluations (Artificial Analysis)Coding & agents · Terminal-Bench 4.04.0%
Long context · AA-LCR82.0%
Knowledge · Humanity's Last Exam47.0%
Instruction following · IFBench77.1%
Sourced Benchmark Evaluations (21)
Verified performance across authoritative benchmarks with provenance tracking.| Benchmark | Status | Score | Trust | Date | Source Type | Provenance |
|---|---|---|---|---|---|---|
| AA-LCR (Long Context Reasoning) | Active | 82% | 93 | 2026-11-15 | independent | Source ↗ |
| APEX-Agents-AA | Active | 32% | 93 | 2026-11-15 | independent | Source ↗ |
| ITBench-AA | Active | 30.3% | 93 | 2026-11-15 | independent | Source ↗ |
| EnterpriseOps-Gym-AA | Active | 42.2% | 93 | 2026-11-15 | independent | Source ↗ |
| Harvey LAB-AA | Active | 0% | 93 | 2026-11-15 | independent | Source ↗ |
| AA-AnalystAgent | Active | 41.3% | 93 | 2026-11-15 | independent | Source ↗ |
| AutomationBench-AA | Active | 35.4% | 93 | 2026-11-15 | independent | Source ↗ |
| MLCR-AA (Medical Long Context Reasoning) | Active | 11.1% | 93 | 2026-11-15 | independent | Source ↗ |
| GDP.pdf | Active | 17.8% | 93 | 2026-11-15 | independent | Source ↗ |
| HLE-Diamond | Active | 31.4% | 93 | 2026-11-15 | independent | Source ↗ |
| CritPt | Active | 17.7% | 93 | 2026-11-15 | independent | Source ↗ |
| GPQA | Nearing Saturation | 94.1% | 78 | 2026-11-15 | independent | Source ↗ |
| Terminal-Bench 4.0 | Active | 4% | 93 | 2026-11-15 | independent | Source ↗ |
| Terminal-Bench 2.1 | Active | 73.8% | 93 | 2026-11-15 | independent | Source ↗ |
| Terminal-Bench Hard | Active | 53.8% | 93 | 2026-11-15 | independent | Source ↗ |
| τ³-Banking | Active | 21.4% | 93 | 2026-11-15 | independent | Source ↗ |
| τ²-Bench Telecom | Active | 95.6% | 93 | 2026-11-15 | independent | Source ↗ |
| SciCode | Active | 58.7% | 93 | 2026-11-15 | independent | Source ↗ |
| IFBench | Active | 77.1% | 93 | 2026-11-15 | independent | Source ↗ |
| MMMU-Pro | Active | 82.4% | 93 | 2026-11-15 | independent | Source ↗ |
| Humanity's Last Exam | Active | 47% | 93 | 2026-11-15 | independent | Source ↗ |
OpenAI Compatible API ExecutionAPI Ready
from openai import OpenAI
client = OpenAI()
response = client.chat.completions.create(
model="google-gemini-3-1-pro-preview",
messages=[{"role": "user", "content": "Evaluate multi-step logic problem"}],
temperature=0.2
)
print(response.choices[0].message.content)Evaluation & Sourcing Notes
Scores listed for Gemini 3.1 Pro represent verified evaluations extracted from official research papers, independent evaluation suites (HELM, LMSYS, OpenCompass, LiveBench), and verified audit reports.
All benchmarks marked as saturated or deprecated reflect historical performance where the benchmark no longer provides active discriminative power.