{"model": {"slug": "claude-3-7-sonnet", "name": "Claude 3.7 Sonnet", "vendor": "Anthropic", "model_family": "claude-3-7-sonnet", "parameter_count_b": null, "context_length": 200000, "homepage_url": null, "hf_model_id": null, "openrouter_id": null, "card_summary": "Anthropic Claude 3.7 Sonnet on SWE-bench Verified leaderboards.", "is_rl_checkpoint": false, "score_count": 23, "best_success_rate": 81.0, "input_price_per_million": null, "output_price_per_million": null}, "scores": [{"success_rate": 81.0, "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_original/", "observed_at": null, "harness": {"slug": "hal-generalist-agent", "name": "HAL Generalist Agent", "vendor": "Princeton PLI", "repo_url": null, "openrouter_icon_url": null}, "model": {"slug": "claude-3-7-sonnet", "name": "Claude 3.7 Sonnet", "vendor": "Anthropic"}, "benchmark": {"slug": "hal-reliability-tau-bench-airline-original", "name": "HAL Reliability \u2014 \u03c4-bench Airline (original)"}}, {"success_rate": 66.4, "resolved_count": 332, "total_count": 500, "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": "SWE-bench Verified public submission", "source": "swe_bench_site", "source_url": "https://github.com/SWE-bench/experiments/blob/main/evaluation/verified/20250514_aime_coder/README.md", "observed_at": "2025-05-14T00:00:00+00:00", "harness": {"slug": "aime-coder-v1", "name": "Aime Coder V1", "vendor": "Swe Bench", "repo_url": null, "openrouter_icon_url": null}, "model": {"slug": "claude-3-7-sonnet", "name": "Claude 3.7 Sonnet", "vendor": "Anthropic"}, "benchmark": {"slug": "swe-bench-verified", "name": "SWE-bench Verified"}}, {"success_rate": 62.4, "resolved_count": 312, "total_count": 500, "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": "SWE-bench Verified public submission", "source": "swe_bench_site", "source_url": "https://github.com/SWE-agent/SWE-agent", "observed_at": "2025-02-25T00:00:00+00:00", "harness": {"slug": "swe-agent", "name": "SWE-agent", "vendor": "SWE-agent", "repo_url": null, "openrouter_icon_url": null}, "model": {"slug": "claude-3-7-sonnet", "name": "Claude 3.7 Sonnet", "vendor": "Anthropic"}, "benchmark": {"slug": "swe-bench-verified", "name": "SWE-bench Verified"}}, {"success_rate": 56.6, "resolved_count": 283, "total_count": 500, "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": "SWE-bench Verified public submission", "source": "swe_bench_site", "source_url": "https://gen42.ai", "observed_at": "2025-04-05T00:00:00+00:00", "harness": {"slug": "swe-rizzo", "name": "SWE Rizzo", "vendor": "SWE-Rizzo", "repo_url": null, "openrouter_icon_url": null}, "model": {"slug": "claude-3-7-sonnet", "name": "Claude 3.7 Sonnet", "vendor": "Anthropic"}, "benchmark": {"slug": "swe-bench-verified", "name": "SWE-bench Verified"}}, {"success_rate": 56.36, "resolved_count": null, "total_count": null, "cost_usd": 0.792, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": 165, "is_harnessrl_measured": false, "eval_protocol": "HAL GAIA (HAL) accuracy", "source": "hal_site", "source_url": "https://hal.cs.princeton.edu/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": "claude-3-7-sonnet", "name": "Claude 3.7 Sonnet", "vendor": "Anthropic"}, "benchmark": {"slug": "hal-gaia", "name": "GAIA (HAL)"}}, {"success_rate": 56.0, "resolved_count": null, "total_count": null, "cost_usd": 0.8422, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": 50, "is_harnessrl_measured": false, "eval_protocol": "HAL TAU-bench Airline (HAL) accuracy", "source": "hal_site", "source_url": "https://hal.cs.princeton.edu/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": "claude-3-7-sonnet", "name": "Claude 3.7 Sonnet", "vendor": "Anthropic"}, "benchmark": {"slug": "hal-tau-bench-airline", "name": "TAU-bench Airline (HAL)"}}, {"success_rate": 50.0, "resolved_count": null, "total_count": null, "cost_usd": 8.2182, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": 49, "is_harnessrl_measured": false, "eval_protocol": "HAL SWE-bench Verified Mini (HAL) accuracy", "source": "hal_site", "source_url": "https://hal.cs.princeton.edu/swebench_verified_mini", "observed_at": null, "harness": {"slug": "swe-agent", "name": "SWE-agent", "vendor": "SWE-agent", "repo_url": null, "openrouter_icon_url": null}, "model": {"slug": "claude-3-7-sonnet", "name": "Claude 3.7 Sonnet", "vendor": "Anthropic"}, "benchmark": {"slug": "hal-swe-bench-verified-mini", "name": "SWE-bench Verified Mini (HAL)"}}, {"success_rate": 48.0, "resolved_count": 144, "total_count": 300, "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": "SWE-bench Lite public submission", "source": "swe_bench_site", "source_url": "https://github.com/SWE-agent/SWE-agent", "observed_at": "2025-02-26T00:00:00+00:00", "harness": {"slug": "swe-agent", "name": "SWE-agent", "vendor": "SWE-agent", "repo_url": null, "openrouter_icon_url": null}, "model": {"slug": "claude-3-7-sonnet", "name": "Claude 3.7 Sonnet", "vendor": "Anthropic"}, "benchmark": {"slug": "swe-bench-lite", "name": "SWE-bench Lite"}}, {"success_rate": 44.0, "resolved_count": null, "total_count": null, "cost_usd": 0.309, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": 50, "is_harnessrl_measured": false, "eval_protocol": "HAL TAU-bench Airline (HAL) accuracy", "source": "hal_site", "source_url": "https://hal.cs.princeton.edu/taubench_airline", "observed_at": null, "harness": {"slug": "tau-bench-tool-calling", "name": "TAU-bench Tool Calling", "vendor": null, "repo_url": null, "openrouter_icon_url": null}, "model": {"slug": "claude-3-7-sonnet", "name": "Claude 3.7 Sonnet", "vendor": "Anthropic"}, "benchmark": {"slug": "hal-tau-bench-airline", "name": "TAU-bench Airline (HAL)"}}, {"success_rate": 38.33, "resolved_count": null, "total_count": null, "cost_usd": 3.0883, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": 300, "is_harnessrl_measured": false, "eval_protocol": "HAL Online Mind2Web (HAL) accuracy", "source": "hal_site", "source_url": "https://hal.cs.princeton.edu/online_mind2web", "observed_at": null, "harness": {"slug": "browser-use", "name": "Browser-Use", "vendor": null, "repo_url": null, "openrouter_icon_url": null}, "model": {"slug": "claude-3-7-sonnet", "name": "Claude 3.7 Sonnet", "vendor": "Anthropic"}, "benchmark": {"slug": "hal-online-mind2web", "name": "Online Mind2Web (HAL)"}}, {"success_rate": 36.97, "resolved_count": null, "total_count": null, "cost_usd": 2.5161, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": 165, "is_harnessrl_measured": false, "eval_protocol": "HAL GAIA (HAL) accuracy", "source": "hal_site", "source_url": "https://hal.cs.princeton.edu/gaia", "observed_at": null, "harness": {"slug": "hf-open-deep-research", "name": "HF Open Deep Research", "vendor": null, "repo_url": null, "openrouter_icon_url": null}, "model": {"slug": "claude-3-7-sonnet", "name": "Claude 3.7 Sonnet", "vendor": "Anthropic"}, "benchmark": {"slug": "hal-gaia", "name": "GAIA (HAL)"}}, {"success_rate": 35.56, "resolved_count": null, "total_count": null, "cost_usd": 1.6231, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": 45, "is_harnessrl_measured": false, "eval_protocol": "HAL CORE-Bench Hard (HAL) accuracy", "source": "hal_site", "source_url": "https://hal.cs.princeton.edu/corebench_hard", "observed_at": null, "harness": {"slug": "core-agent", "name": "CORE-Agent", "vendor": null, "repo_url": null, "openrouter_icon_url": null}, "model": {"slug": "claude-3-7-sonnet", "name": "Claude 3.7 Sonnet", "vendor": "Anthropic"}, "benchmark": {"slug": "hal-corebench-hard", "name": "CORE-Bench Hard (HAL)"}}, {"success_rate": 31.33, "resolved_count": 162, "total_count": 517, "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": "SWE-bench Multimodal public submission", "source": "swe_bench_site", "source_url": "https://github.com/adityasoni9998/OpenHands-Versa", "observed_at": "2025-05-09T00:00:00+00:00", "harness": {"slug": "openhands-sdk", "name": "OpenHands SDK", "vendor": "All Hands AI", "repo_url": null, "openrouter_icon_url": null}, "model": {"slug": "claude-3-7-sonnet", "name": "Claude 3.7 Sonnet", "vendor": "Anthropic"}, "benchmark": {"slug": "swe-bench-multimodal-site", "name": "SWE-bench Multimodal (site)"}}, {"success_rate": 31.11, "resolved_count": null, "total_count": null, "cost_usd": 1.2587, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": 45, "is_harnessrl_measured": false, "eval_protocol": "HAL CORE-Bench Hard (HAL) accuracy", "source": "hal_site", "source_url": "https://hal.cs.princeton.edu/corebench_hard", "observed_at": null, "harness": {"slug": "hal-generalist-agent", "name": "HAL Generalist Agent", "vendor": "Princeton PLI", "repo_url": null, "openrouter_icon_url": null}, "model": {"slug": "claude-3-7-sonnet", "name": "Claude 3.7 Sonnet", "vendor": "Anthropic"}, "benchmark": {"slug": "hal-corebench-hard", "name": "CORE-Bench Hard (HAL)"}}, {"success_rate": 29.32, "resolved_count": null, "total_count": null, "cost_usd": 0.2867, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": 135, "is_harnessrl_measured": false, "eval_protocol": "HAL USACO (HAL) accuracy", "source": "hal_site", "source_url": "https://hal.cs.princeton.edu/usaco", "observed_at": null, "harness": {"slug": "usaco-episodic-semantic", "name": "USACO Episodic + Semantic", "vendor": null, "repo_url": null, "openrouter_icon_url": null}, "model": {"slug": "claude-3-7-sonnet", "name": "Claude 3.7 Sonnet", "vendor": "Anthropic"}, "benchmark": {"slug": "hal-usaco", "name": "USACO (HAL)"}}, {"success_rate": 28.33, "resolved_count": null, "total_count": null, "cost_usd": 0.9732, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": 300, "is_harnessrl_measured": false, "eval_protocol": "HAL Online Mind2Web (HAL) accuracy", "source": "hal_site", "source_url": "https://hal.cs.princeton.edu/online_mind2web", "observed_at": null, "harness": {"slug": "seeact", "name": "SeeAct", "vendor": null, "repo_url": null, "openrouter_icon_url": null}, "model": {"slug": "claude-3-7-sonnet", "name": "Claude 3.7 Sonnet", "vendor": "Anthropic"}, "benchmark": {"slug": "hal-online-mind2web", "name": "Online Mind2Web (HAL)"}}, {"success_rate": 26.0, "resolved_count": null, "total_count": null, "cost_usd": 2.3965, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": 49, "is_harnessrl_measured": false, "eval_protocol": "HAL SWE-bench Verified Mini (HAL) accuracy", "source": "hal_site", "source_url": "https://hal.cs.princeton.edu/swebench_verified_mini", "observed_at": null, "harness": {"slug": "hal-generalist-agent", "name": "HAL Generalist Agent", "vendor": "Princeton PLI", "repo_url": null, "openrouter_icon_url": null}, "model": {"slug": "claude-3-7-sonnet", "name": "Claude 3.7 Sonnet", "vendor": "Anthropic"}, "benchmark": {"slug": "hal-swe-bench-verified-mini", "name": "SWE-bench Verified Mini (HAL)"}}, {"success_rate": 22.55, "resolved_count": null, "total_count": null, "cost_usd": 0.0698, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": 102, "is_harnessrl_measured": false, "eval_protocol": "HAL ScienceAgentBench (HAL) accuracy", "source": "hal_site", "source_url": "https://hal.cs.princeton.edu/scienceagentbench", "observed_at": null, "harness": {"slug": "sab-self-debug", "name": "SAB Self Debug", "vendor": null, "repo_url": null, "openrouter_icon_url": null}, "model": {"slug": "claude-3-7-sonnet", "name": "Claude 3.7 Sonnet", "vendor": "Anthropic"}, "benchmark": {"slug": "hal-scienceagentbench", "name": "ScienceAgentBench (HAL)"}}, {"success_rate": 16.69, "resolved_count": null, "total_count": null, "cost_usd": 1.8667, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": 30, "is_harnessrl_measured": false, "eval_protocol": "HAL AssistantBench (HAL) accuracy", "source": "hal_site", "source_url": "https://hal.cs.princeton.edu/assistantbench", "observed_at": null, "harness": {"slug": "browser-use", "name": "Browser-Use", "vendor": null, "repo_url": null, "openrouter_icon_url": null}, "model": {"slug": "claude-3-7-sonnet", "name": "Claude 3.7 Sonnet", "vendor": "Anthropic"}, "benchmark": {"slug": "hal-assistantbench", "name": "AssistantBench (HAL)"}}, {"success_rate": 10.78, "resolved_count": null, "total_count": null, "cost_usd": 0.4041, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": 102, "is_harnessrl_measured": false, "eval_protocol": "HAL ScienceAgentBench (HAL) accuracy", "source": "hal_site", "source_url": "https://hal.cs.princeton.edu/scienceagentbench", "observed_at": null, "harness": {"slug": "hal-generalist-agent", "name": "HAL Generalist Agent", "vendor": "Princeton PLI", "repo_url": null, "openrouter_icon_url": null}, "model": {"slug": "claude-3-7-sonnet", "name": "Claude 3.7 Sonnet", "vendor": "Anthropic"}, "benchmark": {"slug": "hal-scienceagentbench", "name": "ScienceAgentBench (HAL)"}}, {"success_rate": 3.08, "resolved_count": null, "total_count": null, "cost_usd": 2.3926, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": 80, "is_harnessrl_measured": false, "eval_protocol": "HAL SciCode (HAL) accuracy", "source": "hal_site", "source_url": "https://hal.cs.princeton.edu/scicode", "observed_at": null, "harness": {"slug": "scicode-tool-calling-agent", "name": "Scicode Tool Calling Agent", "vendor": null, "repo_url": null, "openrouter_icon_url": null}, "model": {"slug": "claude-3-7-sonnet", "name": "Claude 3.7 Sonnet", "vendor": "Anthropic"}, "benchmark": {"slug": "hal-scicode", "name": "SciCode (HAL)"}}, {"success_rate": 3.08, "resolved_count": null, "total_count": null, "cost_usd": 0.755, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": 80, "is_harnessrl_measured": false, "eval_protocol": "HAL SciCode (HAL) accuracy", "source": "hal_site", "source_url": "https://hal.cs.princeton.edu/scicode", "observed_at": null, "harness": {"slug": "hal-generalist-agent", "name": "HAL Generalist Agent", "vendor": "Princeton PLI", "repo_url": null, "openrouter_icon_url": null}, "model": {"slug": "claude-3-7-sonnet", "name": "Claude 3.7 Sonnet", "vendor": "Anthropic"}, "benchmark": {"slug": "hal-scicode", "name": "SciCode (HAL)"}}, {"success_rate": 0.0, "resolved_count": null, "total_count": null, "cost_usd": 0.0638, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "harness_version": null, "trial_count": 80, "is_harnessrl_measured": false, "eval_protocol": "HAL SciCode (HAL) accuracy", "source": "hal_site", "source_url": "https://hal.cs.princeton.edu/scicode", "observed_at": null, "harness": {"slug": "scicode-zero-shot-agent", "name": "Scicode Zero Shot Agent", "vendor": null, "repo_url": null, "openrouter_icon_url": null}, "model": {"slug": "claude-3-7-sonnet", "name": "Claude 3.7 Sonnet", "vendor": "Anthropic"}, "benchmark": {"slug": "hal-scicode", "name": "SciCode (HAL)"}}], "harnesses": [{"slug": "aime-coder-v1", "name": "Aime Coder V1", "vendor": "Swe Bench", "repo_url": "https://github.com/SWE-bench/experiments/blob/main/evaluation/verified/20250514_aime_coder/README.md", "openrouter_icon_url": null}, {"slug": "browser-use", "name": "Browser-Use", "vendor": null, "repo_url": null, "openrouter_icon_url": null}, {"slug": "core-agent", "name": "CORE-Agent", "vendor": null, "repo_url": null, "openrouter_icon_url": null}, {"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": "hf-open-deep-research", "name": "HF Open Deep Research", "vendor": null, "repo_url": null, "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": "sab-self-debug", "name": "SAB Self Debug", "vendor": null, "repo_url": null, "openrouter_icon_url": null}, {"slug": "swe-rizzo", "name": "SWE Rizzo", "vendor": "SWE-Rizzo", "repo_url": null, "openrouter_icon_url": null}, {"slug": "swe-agent", "name": "SWE-agent", "vendor": "SWE-agent", "repo_url": "https://github.com/SWE-agent/SWE-agent", "openrouter_icon_url": null}, {"slug": "scicode-tool-calling-agent", "name": "Scicode Tool Calling Agent", "vendor": null, "repo_url": null, "openrouter_icon_url": null}, {"slug": "scicode-zero-shot-agent", "name": "Scicode Zero Shot Agent", "vendor": null, "repo_url": null, "openrouter_icon_url": null}, {"slug": "seeact", "name": "SeeAct", "vendor": null, "repo_url": null, "openrouter_icon_url": null}, {"slug": "tau-bench-tool-calling", "name": "TAU-bench Tool Calling", "vendor": null, "repo_url": null, "openrouter_icon_url": null}, {"slug": "usaco-episodic-semantic", "name": "USACO Episodic + Semantic", "vendor": null, "repo_url": null, "openrouter_icon_url": null}]}