{"harness": {"slug": "gemini-cli", "name": "Gemini CLI", "vendor": "Google", "description": "Gemini CLI is Google's open terminal agent for coding workflows, built around Gemini models with tool use, file operations, and shell access. It competes in the same product harness category as Claude Code, Codex, and Cursor CLI.\n\nPublic scores appear on Artificial Analysis coding-agent benchmarks and occasional SWE-bench Verified submissions when Google or community members publish results. Each row's model column shows the exact Gemini variant used.\n\nRepository: `google-gemini/gemini-cli`. This catalog does not host Google's internal eval configs \u2014 only publicly linked leaderboard snapshots.", "card_summary": "Google's Gemini CLI \u2014 terminal agent with tool use for repo editing and shell tasks.", "homepage_url": "https://github.com/google-gemini/gemini-cli", "repo_url": "https://github.com/google-gemini/gemini-cli", "docs_url": null, "scaffold_code": null, "harbor_agent_name": null, "api_protocol": null, "status": "validated", "supports_rl": true, "score_count": 5, "best_success_rate": 76.03, "github_stars": null, "popularity_tier": null, "hrl_score": null, "hrl_rank": null, "catalog_token_volume": null, "openrouter_icon_url": "https://www.gstatic.com/lamda/images/gemini_favicon_f06995846fc77c897+24.png", "leaderboard_icon_url": null, "aa_coding_index": 0.3293046837328304, "aa_mean_cost_usd": 1.0400415618609393, "aa_mean_total_tokens": 4727804}, "scores": [{"success_rate": 76.03, "resolved_count": null, "total_count": null, "cost_usd": 1.0400415618609393, "input_tokens": 2465901, "output_tokens": 12977, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": null, "is_harnessrl_measured": false, "eval_protocol": "Artificial Analysis Coding Agent Index component (pass@1)", "source": "artificialanalysis", "source_url": "https://artificialanalysis.ai/agents/coding-agents", "observed_at": "2026-08-27T21:54:11.189561+00:00", "harness": {"slug": "gemini-cli", "name": "Gemini CLI", "vendor": "Google", "repo_url": "https://github.com/google-gemini/gemini-cli", "openrouter_icon_url": "https://www.gstatic.com/lamda/images/gemini_favicon_f06995846fc77c897+24.png"}, "model": {"slug": "gemini-3-1-pro-high", "name": "Gemini 3.1 Pro (high)", "vendor": "Google"}, "benchmark": {"slug": "terminal-bench-v2-1", "name": "Terminal-Bench v2.1 (AA component)"}}, {"success_rate": 65.84, "resolved_count": null, "total_count": null, "cost_usd": 0.5567640449438203, "input_tokens": 47653156, "output_tokens": 7219388, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": 445, "is_harnessrl_measured": false, "eval_protocol": null, "source": "tbench_official", "source_url": "https://github.com/harbor-framework/terminal-bench-2-1/pull/66", "observed_at": "2026-05-01T00:00:00+00:00", "harness": {"slug": "gemini-cli", "name": "Gemini CLI", "vendor": "Google", "repo_url": "https://github.com/google-gemini/gemini-cli", "openrouter_icon_url": "https://www.gstatic.com/lamda/images/gemini_favicon_f06995846fc77c897+24.png"}, "model": {"slug": "gemini-3-pro", "name": "Gemini 3 Pro", "vendor": "Google"}, "benchmark": {"slug": "terminal-bench-2-1-official", "name": "Terminal-Bench 2.1 (official)"}}, {"success_rate": 65.84, "resolved_count": null, "total_count": null, "cost_usd": 0.5314382022471911, "input_tokens": 44843158, "output_tokens": 6123596, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": 445, "is_harnessrl_measured": false, "eval_protocol": null, "source": "tbench_official", "source_url": "https://github.com/harbor-framework/terminal-bench-2-1/pull/68", "observed_at": "2026-05-05T00:00:00+00:00", "harness": {"slug": "gemini-cli", "name": "Gemini CLI", "vendor": "Google", "repo_url": "https://github.com/google-gemini/gemini-cli", "openrouter_icon_url": "https://www.gstatic.com/lamda/images/gemini_favicon_f06995846fc77c897+24.png"}, "model": {"slug": "gemini-3-1-pro", "name": "Gemini 3.1 Pro", "vendor": "Google"}, "benchmark": {"slug": "terminal-bench-2-1-official", "name": "Terminal-Bench 2.1 (official)"}}, {"success_rate": 14.16, "resolved_count": null, "total_count": null, "cost_usd": 1.0400415618609393, "input_tokens": 2465901, "output_tokens": 12977, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": null, "is_harnessrl_measured": false, "eval_protocol": "Artificial Analysis Coding Agent Index component (pass@1)", "source": "artificialanalysis", "source_url": "https://artificialanalysis.ai/agents/coding-agents", "observed_at": "2026-08-27T21:54:11.189561+00:00", "harness": {"slug": "gemini-cli", "name": "Gemini CLI", "vendor": "Google", "repo_url": "https://github.com/google-gemini/gemini-cli", "openrouter_icon_url": "https://www.gstatic.com/lamda/images/gemini_favicon_f06995846fc77c897+24.png"}, "model": {"slug": "gemini-3-1-pro-high", "name": "Gemini 3.1 Pro (high)", "vendor": "Google"}, "benchmark": {"slug": "deepswe", "name": "DeepSWE (AA component)"}}, {"success_rate": 8.6, "resolved_count": null, "total_count": null, "cost_usd": 1.0400415618609393, "input_tokens": 2465901, "output_tokens": 12977, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": null, "is_harnessrl_measured": false, "eval_protocol": "Artificial Analysis Coding Agent Index component (pass@1)", "source": "artificialanalysis", "source_url": "https://artificialanalysis.ai/agents/coding-agents", "observed_at": "2026-08-27T21:54:11.189561+00:00", "harness": {"slug": "gemini-cli", "name": "Gemini CLI", "vendor": "Google", "repo_url": "https://github.com/google-gemini/gemini-cli", "openrouter_icon_url": "https://www.gstatic.com/lamda/images/gemini_favicon_f06995846fc77c897+24.png"}, "model": {"slug": "gemini-3-1-pro-high", "name": "Gemini 3.1 Pro (high)", "vendor": "Google"}, "benchmark": {"slug": "swe-atlas-qna", "name": "SWE-Atlas-QnA (AA component)"}}], "benchmarks": [{"slug": "deepswe", "name": "DeepSWE (AA component)", "category": "eval"}, {"slug": "swe-atlas-qna", "name": "SWE-Atlas-QnA (AA component)", "category": "eval"}, {"slug": "terminal-bench-2-1-official", "name": "Terminal-Bench 2.1 (official)", "category": "eval"}, {"slug": "terminal-bench-v2-1", "name": "Terminal-Bench v2.1 (AA component)", "category": "eval"}]}