{"benchmark": {"slug": "swe-bench-bash-only", "name": "SWE-bench Verified (bash-only)", "description": "The bash-only board on swebench.com runs every model through the same mini-SWE-agent configuration on SWE-bench Verified (500 tasks). It isolates foundation-model capability from vendor CLI scaffolding.\n\nCatalog rows use harness `mini-swe-agent` and `source = swe_bench_site`. Compare bash-only rows against product harness submissions on the main Verified board only when explaining scaffold effects \u2014 not as interchangeable rankings.\n\nRelease tags on mini-SWE-agent correspond to leaderboard version numbers in experiment folder names.", "category": "eval", "split": "verified", "language": "python", "task_count": 500, "homepage_url": "https://www.swebench.com/verified", "paper_url": "https://arxiv.org/abs/2310.06770", "hf_dataset_id": null, "official_leaderboard_url": "https://www.swebench.com/verified", "is_public_leaderboard": true, "harness_count": 1, "model_count": 46, "score_count": 46, "best_success_rate": 76.8}, "protocol": {"eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "temperature": null, "max_turns": null, "context_budget_tokens": null, "n": 46}, "rankings": {"benchmark": {"slug": "swe-bench-bash-only", "name": "SWE-bench Verified (bash-only)", "description": "The bash-only board on swebench.com runs every model through the same mini-SWE-agent configuration on SWE-bench Verified (500 tasks). It isolates foundation-model capability from vendor CLI scaffolding.\n\nCatalog rows use harness `mini-swe-agent` and `source = swe_bench_site`. Compare bash-only rows against product harness submissions on the main Verified board only when explaining scaffold effects \u2014 not as interchangeable rankings.\n\nRelease tags on mini-SWE-agent correspond to leaderboard version numbers in experiment folder names.", "category": "eval", "split": "verified", "language": "python", "task_count": 500, "homepage_url": "https://www.swebench.com/verified", "paper_url": "https://arxiv.org/abs/2310.06770", "hf_dataset_id": null, "official_leaderboard_url": "https://www.swebench.com/verified", "is_public_leaderboard": true, "harness_count": null, "model_count": null, "score_count": null, "best_success_rate": null}, "sort": "success_rate", "rows": [{"rank": 1, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "claude-4-5-opus-high", "name": "Claude 4.5 Opus (high)", "vendor": "Anthropic"}, "success_rate": 76.8, "resolved_count": 384, "total_count": 500, "cost_usd": 0.753907997, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://www.anthropic.com/claude", "observed_at": "2026-02-17T00:00:00+00:00"}, {"rank": 2, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "gemini-3-flash-high", "name": "Gemini 3 Flash (high)", "vendor": "Google"}, "success_rate": 75.8, "resolved_count": 379, "total_count": 500, "cost_usd": 0.35596413079999994, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://ai.google.dev/gemini-api/docs/models", "observed_at": "2026-02-17T00:00:00+00:00"}, {"rank": 3, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "minimax-m2-5-high", "name": "MiniMax M2.5 (high)", "vendor": "MiniMax"}, "success_rate": 75.8, "resolved_count": 379, "total_count": 500, "cost_usd": 0.073289987844, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://www.minimax.io", "observed_at": "2026-02-17T00:00:00+00:00"}, {"rank": 4, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "claude-4-6-opus", "name": "Claude 4.6 Opus", "vendor": "Anthropic"}, "success_rate": 75.6, "resolved_count": 378, "total_count": 500, "cost_usd": 0.5515226445, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://www.anthropic.com/claude", "observed_at": "2026-02-17T00:00:00+00:00"}, {"rank": 5, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "claude-4-5-opus-medium", "name": "Claude 4.5 Opus (medium)", "vendor": "Anthropic"}, "success_rate": 74.4, "resolved_count": 372, "total_count": 500, "cost_usd": 0.721244997, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://www.anthropic.com/news/claude-opus-4-5", "observed_at": "2025-11-24T00:00:00+00:00"}, {"rank": 6, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "gemini-3-pro-preview", "name": "Gemini 3 Pro Preview", "vendor": "Google"}, "success_rate": 74.2, "resolved_count": 371, "total_count": 500, "cost_usd": 0.45995751895000003, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://ai.google.dev/gemini-api/docs/models", "observed_at": "2025-11-18T00:00:00+00:00"}, {"rank": 7, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "gpt-5-2-codex", "name": "GPT-5.2 codex", "vendor": "openai"}, "success_rate": 72.8, "resolved_count": 364, "total_count": 500, "cost_usd": 0.4494238987, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://platform.openai.com/docs/models/gpt-5", "observed_at": "2026-02-19T00:00:00+00:00"}, {"rank": 8, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "glm-5-high", "name": "GLM 5 (high)", "vendor": "Zhipu"}, "success_rate": 72.8, "resolved_count": 364, "total_count": 500, "cost_usd": 0.53438913622, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://z.ai", "observed_at": "2026-02-17T00:00:00+00:00"}, {"rank": 9, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "gpt-5-2-high", "name": "GPT-5.2 (high)", "vendor": "OpenAI"}, "success_rate": 71.8, "resolved_count": 359, "total_count": 500, "cost_usd": 0.5202590036, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://platform.openai.com/docs/models/gpt-5", "observed_at": "2025-12-11T00:00:00+00:00"}, {"rank": 10, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "claude-4-5-sonnet-high", "name": "Claude 4.5 Sonnet (high)", "vendor": "Anthropic"}, "success_rate": 71.4, "resolved_count": 357, "total_count": 500, "cost_usd": 0.6578964984000001, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://www.anthropic.com/claude", "observed_at": "2026-02-17T00:00:00+00:00"}, {"rank": 11, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "kimi-k2-5-high", "name": "KIMI K2.5 (high)", "vendor": "Moonshot"}, "success_rate": 70.8, "resolved_count": 354, "total_count": 500, "cost_usd": 0.14655374074400002, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://moonshotai.github.io/Kimi-K2/", "observed_at": "2026-02-17T00:00:00+00:00"}, {"rank": 12, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "claude-4-5-sonnet", "name": "Claude 4.5 Sonnet", "vendor": "Anthropic"}, "success_rate": 70.6, "resolved_count": 353, "total_count": 500, "cost_usd": 0.5583347409, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://www.anthropic.com/news/claude-sonnet-4-5", "observed_at": "2025-09-29T00:00:00+00:00"}, {"rank": 13, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "deepseek-v3-2-high", "name": "DeepSeek V3.2 (high)", "vendor": "DeepSeek"}, "success_rate": 70.0, "resolved_count": 350, "total_count": 500, "cost_usd": 0.4478464965, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://api-docs.deepseek.com", "observed_at": "2026-02-17T00:00:00+00:00"}, {"rank": 14, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "gemini-3-pro", "name": "Gemini 3 Pro", "vendor": "Google"}, "success_rate": 69.6, "resolved_count": 348, "total_count": 500, "cost_usd": 0.9600198564, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://ai.google.dev/gemini-api/docs/models", "observed_at": "2026-02-26T00:00:00+00:00"}, {"rank": 15, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "gpt-5-2", "name": "GPT-5.2", "vendor": "openai"}, "success_rate": 69.0, "resolved_count": 345, "total_count": 500, "cost_usd": 0.2696674974, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://platform.openai.com/docs/models/", "observed_at": "2025-12-11T00:00:00+00:00"}, {"rank": 16, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "claude-4-opus", "name": "Claude 4 Opus", "vendor": "Anthropic"}, "success_rate": 67.6, "resolved_count": 338, "total_count": 500, "cost_usd": 1.131270447, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://www.anthropic.com/news/claude-4", "observed_at": "2025-08-02T00:00:00+00:00"}, {"rank": 17, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "claude-4-5-haiku-high", "name": "Claude 4.5 Haiku (high)", "vendor": "Anthropic"}, "success_rate": 66.6, "resolved_count": 333, "total_count": 500, "cost_usd": 0.33092443610000005, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://www.anthropic.com/claude", "observed_at": "2026-02-17T00:00:00+00:00"}, {"rank": 18, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "gpt-5-1-codex-medium", "name": "GPT-5.1 codex (medium)", "vendor": "OpenAI"}, "success_rate": 66.0, "resolved_count": 330, "total_count": 500, "cost_usd": 0.5888823500000001, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://platform.openai.com/docs/models/gpt-5", "observed_at": "2025-11-24T00:00:00+00:00"}, {"rank": 19, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "gpt-5-1-medium", "name": "GPT-5.1 (medium)", "vendor": "OpenAI"}, "success_rate": 66.0, "resolved_count": 330, "total_count": 500, "cost_usd": 0.3061554725, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://platform.openai.com/docs/models/gpt-5", "observed_at": "2025-11-20T00:00:00+00:00"}, {"rank": 20, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "gpt-5-medium", "name": "GPT-5 (medium)", "vendor": "OpenAI"}, "success_rate": 65.0, "resolved_count": 325, "total_count": 500, "cost_usd": 0.2803830175, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://platform.openai.com/docs/models/gpt-5", "observed_at": "2025-08-07T00:00:00+00:00"}, {"rank": 21, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "claude-4-sonnet", "name": "Claude 4 Sonnet", "vendor": "Anthropic"}, "success_rate": 64.93, "resolved_count": 325, "total_count": 500, "cost_usd": 0.37145316780000004, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://www.anthropic.com/news/claude-4", "observed_at": "2025-07-26T00:00:00+00:00"}, {"rank": 22, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "kimi-k2-thinking", "name": "KIMI K2 Thinking", "vendor": "moonshotai"}, "success_rate": 63.4, "resolved_count": 317, "total_count": 500, "cost_usd": 0.4383072354, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://moonshotai.github.io/Kimi-K2/", "observed_at": "2025-12-10T00:00:00+00:00"}, {"rank": 23, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "minimax-m2", "name": "MiniMax M2", "vendor": "minimax"}, "success_rate": 61.0, "resolved_count": 305, "total_count": 500, "cost_usd": 0.42825617729, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://www.minimax.io", "observed_at": "2025-11-24T00:00:00+00:00"}, {"rank": 24, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "deepseek-v3-2-reasoner", "name": "DeepSeek V3.2 Reasoner", "vendor": "DeepSeek"}, "success_rate": 60.0, "resolved_count": 300, "total_count": 500, "cost_usd": 0.0280763028, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://www.swebench.com/verified", "observed_at": "2025-12-01T00:00:00+00:00"}, {"rank": 25, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "gpt-5-mini-medium", "name": "GPT-5 mini (medium)", "vendor": "OpenAI"}, "success_rate": 59.8, "resolved_count": 299, "total_count": 500, "cost_usd": 0.035477067300000005, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://platform.openai.com/docs/models/gpt-5-mini", "observed_at": "2025-08-07T00:00:00+00:00"}, {"rank": 26, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "o3", "name": "o3", "vendor": "openai"}, "success_rate": 58.4, "resolved_count": 292, "total_count": 500, "cost_usd": 0.333652748, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://platform.openai.com/docs/models/o3", "observed_at": "2025-07-26T00:00:00+00:00"}, {"rank": 27, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "devstral-small-2512", "name": "Devstral Small (2512)", "vendor": "Mistral"}, "success_rate": 56.4, "resolved_count": 282, "total_count": 500, "cost_usd": null, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://www.swebench.com/verified", "observed_at": "2025-12-09T00:00:00+00:00"}, {"rank": 28, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "gpt-5-mini", "name": "GPT-5 mini", "vendor": "openai"}, "success_rate": 56.2, "resolved_count": 281, "total_count": 500, "cost_usd": 0.0472012191, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://platform.openai.com", "observed_at": "2026-02-17T00:00:00+00:00"}, {"rank": 29, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "qwen3-coder-480b-a35b-instruct", "name": "Qwen3 Coder 480B A35B Instruct", "vendor": "Qwen"}, "success_rate": 55.4, "resolved_count": 277, "total_count": 500, "cost_usd": 0.247927108, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://qwenlm.github.io/blog/qwen3-coder/", "observed_at": "2025-08-02T00:00:00+00:00"}, {"rank": 30, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "glm-4-6-t=1", "name": "GLM 4.6 (T=1)", "vendor": "Zhipu"}, "success_rate": 55.4, "resolved_count": 277, "total_count": 500, "cost_usd": 0.09661007519999999, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://www.swebench.com/verified", "observed_at": "2025-12-01T00:00:00+00:00"}, {"rank": 31, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "glm-4-5-2025-08-22", "name": "GLM 4.5 (2025 08 22)", "vendor": "Zhipu"}, "success_rate": 54.2, "resolved_count": 271, "total_count": 500, "cost_usd": 0.2971903488, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://z.ai/blog/glm-4.5", "observed_at": "2025-08-22T00:00:00+00:00"}, {"rank": 32, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "devstral-2512", "name": "Devstral (2512)", "vendor": "mistralai"}, "success_rate": 53.8, "resolved_count": 269, "total_count": 500, "cost_usd": null, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://www.swebench.com/verified", "observed_at": "2025-12-09T00:00:00+00:00"}, {"rank": 33, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "gemini-2-5-pro-20250506", "name": "Gemini 2.5 Pro (2025-05-06)", "vendor": "Google"}, "success_rate": 53.6, "resolved_count": 268, "total_count": 500, "cost_usd": 0.28837209562500005, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://ai.google.dev/gemini-api/docs/models", "observed_at": "2025-07-26T00:00:00+00:00"}, {"rank": 34, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "claude-3-7-sonnet-20250219", "name": "Claude 3.7 Sonnet (2025-02-19)", "vendor": "Anthropic"}, "success_rate": 52.8, "resolved_count": 264, "total_count": 500, "cost_usd": 0.3546154482, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://www.anthropic.com/news/claude-3-7-sonnet", "observed_at": "2025-07-20T00:00:00+00:00"}, {"rank": 35, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "o4-mini", "name": "o4 mini", "vendor": "openai"}, "success_rate": 45.0, "resolved_count": 225, "total_count": 500, "cost_usd": 0.2099733328, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://platform.openai.com/docs/models/o4-mini", "observed_at": "2025-07-26T00:00:00+00:00"}, {"rank": 36, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "kimi-k2-instruct", "name": "KIMI K2 Instruct", "vendor": "Moonshot"}, "success_rate": 43.8, "resolved_count": 219, "total_count": 500, "cost_usd": 0.531593366, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://moonshotai.github.io/Kimi-K2/", "observed_at": "2025-08-07T00:00:00+00:00"}, {"rank": 37, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "gpt-4-1-20250414", "name": "GPT-4.1 (2025-04-14)", "vendor": "OpenAI"}, "success_rate": 39.58, "resolved_count": 198, "total_count": 500, "cost_usd": 0.146313124, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://platform.openai.com/docs/models/gpt-4.1", "observed_at": "2025-07-26T00:00:00+00:00"}, {"rank": 38, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "gpt-5-nano-medium", "name": "GPT-5 nano (medium)", "vendor": "OpenAI"}, "success_rate": 34.8, "resolved_count": 174, "total_count": 500, "cost_usd": 0.03807518078, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://platform.openai.com/docs/models/gpt-5-nano", "observed_at": "2025-08-07T00:00:00+00:00"}, {"rank": 39, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "gemini-2-5-flash-20250417", "name": "Gemini 2.5 Flash (2025-04-17)", "vendor": "Google"}, "success_rate": 28.73, "resolved_count": 144, "total_count": 500, "cost_usd": 0.13388571835, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://ai.google.dev/gemini-api/docs/models", "observed_at": "2025-07-26T00:00:00+00:00"}, {"rank": 40, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "gpt-oss-120b", "name": "GPT Oss 120b", "vendor": "openai"}, "success_rate": 26.0, "resolved_count": 130, "total_count": 500, "cost_usd": 0.05711905579999999, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://platform.openai.com/docs/models/gpt-oss-120b", "observed_at": "2025-08-07T00:00:00+00:00"}, {"rank": 41, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "gpt-4-1-mini-20250414", "name": "GPT-4.1 mini (2025-04-14)", "vendor": "OpenAI"}, "success_rate": 23.94, "resolved_count": 120, "total_count": 500, "cost_usd": 0.43515415279999997, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://platform.openai.com/docs/models/gpt-4.1-mini", "observed_at": "2025-07-20T00:00:00+00:00"}, {"rank": 42, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "gpt-4o-20241120", "name": "GPT-4o (2024-11-20)", "vendor": "OpenAI"}, "success_rate": 21.62, "resolved_count": 108, "total_count": 500, "cost_usd": 1.531148125, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://platform.openai.com/docs/models/gpt-4o", "observed_at": "2025-07-20T00:00:00+00:00"}, {"rank": 43, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "llama-4-maverick-instruct", "name": "Llama 4 Maverick Instruct", "vendor": "Meta"}, "success_rate": 21.04, "resolved_count": 105, "total_count": 500, "cost_usd": 0.31370609759999996, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://www.together.ai/models/llama-4-maverick", "observed_at": "2025-07-20T00:00:00+00:00"}, {"rank": 44, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "gemini-2-0-flash", "name": "Gemini 2.0 Flash", "vendor": "Google"}, "success_rate": 13.52, "resolved_count": 68, "total_count": 500, "cost_usd": 0.0, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://ai.google.dev/gemini-api/docs/models", "observed_at": "2025-07-26T00:00:00+00:00"}, {"rank": 45, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "llama-4-scout-instruct", "name": "Llama 4 Scout Instruct", "vendor": "Meta"}, "success_rate": 9.06, "resolved_count": 45, "total_count": 500, "cost_usd": 0.115631523, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://www.together.ai/models/llama-4-scout", "observed_at": "2025-07-20T00:00:00+00:00"}, {"rank": 46, "harness": {"slug": "mini-swe-agent", "name": "mini-SWE-agent", "vendor": "anthropic", "scaffold_code": null, "repo_url": "https://github.com/SWE-agent/mini-swe-agent", "homepage_url": "https://github.com/SWE-agent/mini-swe-agent", "openrouter_icon_url": null, "leaderboard_icon_url": "https://www.swebench.com/img/mini-icon.svg"}, "model": {"slug": "qwen2-5-coder-32b-instruct", "name": "Qwen2.5 Coder 32B Instruct", "vendor": "qwen"}, "success_rate": 9.0, "resolved_count": 45, "total_count": 500, "cost_usd": 0.06811614065999999, "input_tokens": null, "output_tokens": null, "latency_ms": null, "temperature": null, "max_turns": null, "context_budget_tokens": null, "is_harnessrl_measured": false, "eval_protocol": "SWE-bench bash-only (mini-SWE-agent)", "source": "swe_bench_site", "source_url": "https://qwenlm.github.io/blog/qwen2.5-coder/", "observed_at": "2025-08-03T00:00:00+00:00"}]}}