Gemini 1.0 Pro: benchmark results
Google's mid-size Gemini 1.0 tier (announced December 2023) that powered Bard and launched the Gemini API, sitting between Nano and Ultra. Provider: Google. Released 2024-02-01. Access: API.
Aggregate Rating 1460 ± 8, rank #933 of 1656 rated models, from 355 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| CyberSecEval-3 Insecure Instruction | 32 | Insecure code generation (%) | 100 |
| EvalYaks CEFR B2 (With Descriptors) | 82 | Average acceptable accuracy (%) across speaking-test parts 1 | 100 |
| EvalYaks CEFR B2 (Without Descriptors) | 69 | Average acceptable accuracy (%) across speaking-test parts 1 | 100 |
| LLM Trustworthy - Adversarial | 67.28 | Trust Score (%) | 97.9 |
| LLM Trustworthy - Ethics | 93.74 | Trust Score (%) | 97.9 |
| OpenVLM MME - Celebrity | 169.4 | Score | 91.5 |
| CyberSecEval-3 Insecure Autocomplete | 25 | Insecure code generation (%) | 90 |
| AI-Secure LLM Trustworthy Leaderboard | 0.81 | Trustworthy average | 89.6 |
| LLM Trustworthy Leaderboard | 80.61 | Average Trust Score (%) | 89.6 |
| Not All Proofs Are Equal: Evaluating LLM Proof Quality Beyond Correctness | 60.7 | Accuracy (%) | 88.9 |
| OpenVLM MMMU - Geography | 63.3 | Accuracy (%) | 88.9 |
| AgentHarm | 7.4 | Harm score (%) | 88.2 |
Interactive version: theaggregate.ai/model?slug=gemini-1-0-pro · How It Works · Data refreshed daily, snapshot 2026-10-20.