✨ Add Opus 4.8

results/swe-bench-pro--ansible-claude-opus-4-8-claude-code.json

@@ -0,0 +1,60 @@
+{
+    "benchmark": {
+        "name": "swe-bench-pro--ansible",
+        "repo": "ScaleAI/SWE-bench_Pro",
+        "num_tasks": 96,
+        "url": "https://huggingface.co/datasets/ScaleAI/SWE-bench_Pro"
+    },
+    "harness": {
+        "name": "Claude Code",
+        "skills": [],
+        "is_oss": false,
+        "url": "https://github.com/anthropics/claude-code"
+    },
+    "model": {
+        "name": "Opus 4.8",
+        "repo": "Opus 4.8",
+        "is_oss": true,
+        "num_params": 1000,
+        "precision": "bf16",
+        "url": "https://www.anthropic.com/news/claude-opus-4-8"
+    },
+    "environment": {
+        "name": "harbor",
+        "config": {
+            "path": null,
+            "name": "scale-ai/swe-bench-pro",
+            "version": null,
+            "ref": "sha256:88411d32ff27e53a4c1a7e29f0c2aeba180c8e5d60f221cab5ed56325f33549d",
+            "registry_url": null,
+            "registry_path": null,
+            "overwrite": false,
+            "download_dir": null,
+            "task_names": [
+                "*ansible*"
+            ],
+            "exclude_task_names": null,
+            "n_tasks": null
+        },
+        "url": "https://github.com/harbor-framework/harbor"
+    },
+    "metrics": {
+        "n_tasks": 96,
+        "n_errors": 0,
+        "score": 0.698,
+        "n_input_tokens": 192346997,
+        "n_cache_tokens": 186506482,
+        "n_output_tokens": 2179945,
+        "n_total_tokens": 381033424,
+        "agent_time_seconds": 32745,
+        "total_time_seconds": 39030,
+        "cost_usd": 185.66155285,
+        "mean_input_tokens_per_task": 2003614,
+        "mean_cache_tokens_per_task": 1942775,
+        "mean_output_tokens_per_task": 22707,
+        "mean_tokens_per_task": 3969098,
+        "mean_cost_usd_per_task": 1.93,
+        "mean_total_time_seconds_per_task": 406,
+        "mean_agent_time_seconds_per_task": 341
+    }
+}

results/swe-bench-verified-claude-opus-4-8-claude-code.json

@@ -0,0 +1,58 @@
+{
+    "benchmark": {
+        "name": "swe-bench-verified",
+        "repo": "SWE-bench/SWE-bench_Verified",
+        "num_tasks": 500,
+        "url": "https://huggingface.co/datasets/SWE-bench/SWE-bench_Verified"
+    },
+    "harness": {
+        "name": "Claude Code",
+        "skills": [],
+        "is_oss": false,
+        "url": "https://github.com/anthropics/claude-code"
+    },
+    "model": {
+        "name": "Opus 4.8",
+        "repo": "Opus 4.8",
+        "is_oss": true,
+        "num_params": 1000,
+        "precision": "bf16",
+        "url": "https://www.anthropic.com/news/claude-opus-4-8"
+    },
+    "environment": {
+        "name": "harbor",
+        "config": {
+            "path": null,
+            "name": "swe-bench/swe-bench-verified",
+            "version": null,
+            "ref": "sha256:235d6032d549851a936db3b5fe08807c4d385c12ee10e7be9c9786a1ff60563c",
+            "registry_url": null,
+            "registry_path": null,
+            "overwrite": false,
+            "download_dir": null,
+            "task_names": null,
+            "exclude_task_names": null,
+            "n_tasks": null
+        },
+        "url": "https://github.com/harbor-framework/harbor"
+    },
+    "metrics": {
+        "n_tasks": 500,
+        "n_errors": 0,
+        "score": 0.868,
+        "n_input_tokens": 402225158,
+        "n_cache_tokens": 388781382,
+        "n_output_tokens": 4588859,
+        "n_total_tokens": 795595399,
+        "agent_time_seconds": 104467,
+        "total_time_seconds": 159566,
+        "cost_usd": 394.7823929,
+        "mean_input_tokens_per_task": 804450,
+        "mean_cache_tokens_per_task": 777562,
+        "mean_output_tokens_per_task": 9177,
+        "mean_tokens_per_task": 1591190,
+        "mean_cost_usd_per_task": 0.79,
+        "mean_total_time_seconds_per_task": 319,
+        "mean_agent_time_seconds_per_task": 208
+    }
+}