{"model": {"slug": "claude-4-sonnet-high", "name": "Claude 4 Sonnet (high)", "vendor": "Anthropic", "model_family": "claude-4-sonnet", "parameter_count_b": null, "context_length": 200000, "homepage_url": null, "hf_model_id": null, "openrouter_id": null, "card_summary": "Anthropic Claude 4 Sonnet (high) on SWE-bench Verified leaderboards.", "is_rl_checkpoint": false, "score_count": 3, "best_success_rate": 39.33, "input_price_per_million": null, "output_price_per_million": null}, "scores": [{"success_rate": 39.33, "resolved_count": null, "total_count": null, "cost_usd": 5.3664, "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-4-sonnet-high", "name": "Claude 4 Sonnet (high)", "vendor": "Anthropic"}, "benchmark": {"slug": "hal-online-mind2web", "name": "Online Mind2Web (HAL)"}}, {"success_rate": 36.67, "resolved_count": null, "total_count": null, "cost_usd": 1.088, "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-4-sonnet-high", "name": "Claude 4 Sonnet (high)", "vendor": "Anthropic"}, "benchmark": {"slug": "hal-online-mind2web", "name": "Online Mind2Web (HAL)"}}, {"success_rate": 33.33, "resolved_count": null, "total_count": null, "cost_usd": 2.2329, "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-4-sonnet-high", "name": "Claude 4 Sonnet (high)", "vendor": "Anthropic"}, "benchmark": {"slug": "hal-corebench-hard", "name": "CORE-Bench Hard (HAL)"}}], "harnesses": [{"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": "seeact", "name": "SeeAct", "vendor": null, "repo_url": null, "openrouter_icon_url": null}]}