{"model": {"slug": "gemini-3-1-pro", "name": "Gemini 3.1 Pro", "vendor": "Google", "model_family": "gemini-3-1-pro", "parameter_count_b": null, "context_length": 1000000, "homepage_url": null, "hf_model_id": null, "openrouter_id": null, "card_summary": "Google Gemini 3.1 Pro on SWE-bench Verified leaderboards.", "is_rl_checkpoint": false, "score_count": 10, "best_success_rate": 90.0, "input_price_per_million": null, "output_price_per_million": null}, "scores": [{"success_rate": 90.0, "resolved_count": null, "total_count": null, "cost_usd": null, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": null, "is_harnessrl_measured": false, "eval_protocol": "HAL reliability composite (mean of 4 dimension aggregates)", "source": "hal_reliability_site", "source_url": "https://hal.cs.princeton.edu/reliability/benchmark/taubench_airline/", "observed_at": null, "harness": {"slug": "hal-generalist-agent", "name": "HAL Generalist Agent", "vendor": "Princeton PLI", "repo_url": null, "openrouter_icon_url": null}, "model": {"slug": "gemini-3-1-pro", "name": "Gemini 3.1 Pro", "vendor": "Google"}, "benchmark": {"slug": "hal-reliability-tau-bench-airline", "name": "HAL Reliability \u2014 \u03c4-bench Airline (clean)"}}, {"success_rate": 89.75, "resolved_count": null, "total_count": null, "cost_usd": null, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": null, "is_harnessrl_measured": false, "eval_protocol": "HAL reliability composite (mean of 4 dimension aggregates)", "source": "hal_reliability_site", "source_url": "https://hal.cs.princeton.edu/reliability/benchmark/gaia/", "observed_at": null, "harness": {"slug": "hal-generalist-agent", "name": "HAL Generalist Agent", "vendor": "Princeton PLI", "repo_url": null, "openrouter_icon_url": null}, "model": {"slug": "gemini-3-1-pro", "name": "Gemini 3.1 Pro", "vendor": "Google"}, "benchmark": {"slug": "hal-reliability-gaia", "name": "HAL Reliability \u2014 GAIA"}}, {"success_rate": 81.8, "resolved_count": null, "total_count": null, "cost_usd": 1.8, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": null, "is_harnessrl_measured": false, "eval_protocol": "OpenHands SDK v1.11.5", "source": "openhands_index", "source_url": "https://laminar.sh/shared/evals/3a131e0e-fdcd-4898-bfeb-7320496bbafd", "observed_at": "2026-07-31T11:38:58.577710", "harness": {"slug": "openhands-sdk", "name": "OpenHands SDK", "vendor": "All Hands AI", "repo_url": null, "openrouter_icon_url": null}, "model": {"slug": "gemini-3-1-pro", "name": "Gemini 3.1 Pro", "vendor": "Google"}, "benchmark": {"slug": "gaia", "name": "GAIA"}}, {"success_rate": 76.8, "resolved_count": null, "total_count": null, "cost_usd": 1.87, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": null, "is_harnessrl_measured": false, "eval_protocol": "OpenHands SDK v1.11.5", "source": "openhands_index", "source_url": "https://laminar.sh/shared/evals/1e2fadc9-bfce-4a2d-80a2-da24b6e707ad", "observed_at": "2026-07-31T11:38:58.577710", "harness": {"slug": "openhands-sdk", "name": "OpenHands SDK", "vendor": "All Hands AI", "repo_url": null, "openrouter_icon_url": null}, "model": {"slug": "gemini-3-1-pro", "name": "Gemini 3.1 Pro", "vendor": "Google"}, "benchmark": {"slug": "openhands-swe-bench", "name": "SWE-bench (OpenHands Index)"}}, {"success_rate": 75.1, "resolved_count": null, "total_count": null, "cost_usd": 1.62, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": null, "is_harnessrl_measured": false, "eval_protocol": "OpenHands SDK v1.11.5", "source": "openhands_index", "source_url": "https://laminar.sh/shared/evals/c44c25a6-214d-4f87-aa3c-fdb8c3531d13", "observed_at": "2026-07-31T11:38:58.577710", "harness": {"slug": "openhands-sdk", "name": "OpenHands SDK", "vendor": "All Hands AI", "repo_url": null, "openrouter_icon_url": null}, "model": {"slug": "gemini-3-1-pro", "name": "Gemini 3.1 Pro", "vendor": "Google"}, "benchmark": {"slug": "swt-bench", "name": "SWT-Bench"}}, {"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": null, "openrouter_icon_url": null}, "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": 65.62, "resolved_count": null, "total_count": null, "cost_usd": 0.5168314606741573, "input_tokens": 27525560, "output_tokens": 12935831, "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/69", "observed_at": "2026-05-05T00:00:00+00:00", "harness": {"slug": "terminus-2", "name": "Terminus 2", "vendor": "Harbor", "repo_url": null, "openrouter_icon_url": null}, "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": 44.1, "resolved_count": null, "total_count": null, "cost_usd": 1.24, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": null, "is_harnessrl_measured": false, "eval_protocol": "OpenHands SDK v1.11.5", "source": "openhands_index", "source_url": "https://laminar.sh/shared/evals/cca1e142-0f1b-45ef-95fc-fde8e4c9d3aa", "observed_at": "2026-07-31T11:38:58.577710", "harness": {"slug": "openhands-sdk", "name": "OpenHands SDK", "vendor": "All Hands AI", "repo_url": null, "openrouter_icon_url": null}, "model": {"slug": "gemini-3-1-pro", "name": "Gemini 3.1 Pro", "vendor": "Google"}, "benchmark": {"slug": "swe-bench-multimodal", "name": "SWE-bench Multimodal (OpenHands Index)"}}, {"success_rate": 40.8, "resolved_count": null, "total_count": null, "cost_usd": 0.3, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": 60, "is_harnessrl_measured": false, "eval_protocol": "WildClawBench overall score (60 tasks)", "source": "wildclawbench_site", "source_url": "https://github.com/internlm/WildClawBench", "observed_at": null, "harness": {"slug": "openclaw", "name": "OpenClaw", "vendor": "OpenClaw", "repo_url": null, "openrouter_icon_url": null}, "model": {"slug": "gemini-3-1-pro", "name": "Gemini 3.1 Pro", "vendor": "Google"}, "benchmark": {"slug": "wildclawbench", "name": "WildClawBench"}}, {"success_rate": 25.0, "resolved_count": null, "total_count": null, "cost_usd": 1.18, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": null, "is_harnessrl_measured": false, "eval_protocol": "OpenHands SDK v1.11.5", "source": "openhands_index", "source_url": "https://laminar.sh/shared/evals/3817b8ac-4a44-42f7-b6ce-08f747db0061", "observed_at": "2026-07-31T11:38:58.577710", "harness": {"slug": "openhands-sdk", "name": "OpenHands SDK", "vendor": "All Hands AI", "repo_url": null, "openrouter_icon_url": null}, "model": {"slug": "gemini-3-1-pro", "name": "Gemini 3.1 Pro", "vendor": "Google"}, "benchmark": {"slug": "commit0", "name": "Commit0"}}], "harnesses": [{"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"}, {"slug": "hal-generalist-agent", "name": "HAL Generalist Agent", "vendor": "Princeton PLI", "repo_url": "https://github.com/princeton-pli/hal-harness", "openrouter_icon_url": null}, {"slug": "openclaw", "name": "OpenClaw", "vendor": "OpenClaw", "repo_url": "https://github.com/openclaw/openclaw", "openrouter_icon_url": null}, {"slug": "openhands-sdk", "name": "OpenHands SDK", "vendor": "All Hands AI", "repo_url": "https://github.com/All-Hands-AI/OpenHands", "openrouter_icon_url": "https://www.all-hands.dev/favicon.ico"}, {"slug": "terminus-2", "name": "Terminus 2", "vendor": "Harbor", "repo_url": "https://github.com/Elvin-Yiming-Du/harbor", "openrouter_icon_url": null}]}