FrontierCode
Would a repository maintainer merge this software change?
Top rankings
One row per model, using its best reported score across effort settings.
| Rank | Model | Best score | Best reported setting |
|---|---|---|---|
| 1 | Claude Opus 5.5Anthropic | 54.64% | medium effortCognition leaderboard6 values · 2 reportsSep 23, 2026 · source |
| 2 | Claude Fable 5Anthropic | 53.48% | xhigh effortCognition leaderboard5 values · 1 reportSep 23, 2026 · source |
| 3 | Claude Opus 5Anthropic | 53.38% | medium effortCognition leaderboard6 values · 2 reportsSep 23, 2026 · source |
| 4 | GPT-6 AstraOpenAI | 53.30% | Reported configurationAnthropic report6 values · 2 reportsSep 22, 2026 · source |
| 5 | Claude Fable 5.1Anthropic | 50.91% | medium effortCognition leaderboard6 values · 2 reportsSep 23, 2026 · source |
| 6 | SWE-2Cognition | 50.00% | max effortCognition leaderboard3 values · 1 reportSep 23, 2026 · source |
| 7 | GPT-6 SolOpenAI | 49.27% | max effortCognition leaderboard5 values · 1 reportSep 23, 2026 · source |
| 8 | Grok 4.6SpaceXAI | 48.01% | high effortCognition leaderboard3 values · 1 reportSep 23, 2026 · source |
| 9 | Grok 4.7SpaceXAI | 47.59% | high effortCognition leaderboard2 values · 1 reportSep 23, 2026 · source |
| 10 | GPT-5.6 SolOpenAI | 47.50% | Reported configurationAnthropic report6 values · 2 reportsSep 22, 2026 · source |
| 11 | Claude Opus 4.8Anthropic | 46.50% | max effortCognition leaderboard5 values · 1 reportSep 23, 2026 · source |
| 12 | Kimi K3Moonshot AI | 44.17% | none effortCognition leaderboard1 value · 1 reportSep 23, 2026 · source |
Effort curve
Every sourced cost-linked effort value for this exact version. Lines connect complete sweeps only.
Definition and comparison boundarypublic methodology
Would a repository maintainer merge this software change?
End-to-end code quality, including correctness, test quality, scope discipline, style, and adherence to repository standards. Higher is better. Cognition's Score is a weighted aggregate of rubric items; solutions that fail a blocking criterion receive 0.
Weighted Score is not all-blocker Pass Rate. Harness and effort vary by row; compare only exact contexts and splits. The feed does not publish attempt counts.
Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Fable 5"; harness "claude-code"; source effort "high"; Score raw source fraction 0.5273; unfair-internet flagged rate 0.002 source fraction (0.2%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Fable 5"; harness "claude-code"; source effort "low"; Score raw source fraction 0.4801; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Fable 5"; harness "claude-code"; source effort "max"; Score raw source fraction 0.5157; unfair-internet flagged rate 0.002 source fraction (0.2%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Fable 5"; harness "claude-code"; source effort "medium"; Score raw source fraction 0.4982; unfair-internet flagged rate 0.003 source fraction (0.3%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Fable 5"; harness "claude-code"; source effort "xhigh"; Score raw source fraction 0.5348; unfair-internet flagged rate 0.003 source fraction (0.3%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Fable 5.1"; harness "claude-code"; source effort "high"; Score raw source fraction 0.5034; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Fable 5.1"; harness "claude-code"; source effort "low"; Score raw source fraction 0.4982; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Fable 5.1"; harness "claude-code"; source effort "max"; Score raw source fraction 0.5028; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Fable 5.1"; harness "claude-code"; source effort "medium"; Score raw source fraction 0.5091; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Fable 5.1"; harness "claude-code"; source effort "xhigh"; Score raw source fraction 0.4873; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Opus 4.6"; harness "claude-code"; source effort "high"; Score raw source fraction 0.2664; unfair-internet flagged rate 0.001 source fraction (0.1%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Opus 4.6"; harness "claude-code"; source effort "low"; Score raw source fraction 0.1831; unfair-internet flagged rate 0.003 source fraction (0.3%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Opus 4.6"; harness "claude-code"; source effort "medium"; Score raw source fraction 0.2367; unfair-internet flagged rate 0.001 source fraction (0.1%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Opus 4.7"; harness "claude-code"; source effort "high"; Score raw source fraction 0.3517; unfair-internet flagged rate 0.001 source fraction (0.1%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Opus 4.7"; harness "claude-code"; source effort "low"; Score raw source fraction 0.2761; unfair-internet flagged rate 0.001 source fraction (0.1%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Opus 4.7"; harness "claude-code"; source effort "max"; Score raw source fraction 0.3854; unfair-internet flagged rate 0.001 source fraction (0.1%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Opus 4.7"; harness "claude-code"; source effort "medium"; Score raw source fraction 0.2891; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Opus 4.7"; harness "claude-code"; source effort "xhigh"; Score raw source fraction 0.3491; unfair-internet flagged rate 0.004 source fraction (0.4%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Opus 4.8"; harness "claude-code"; source effort "high"; Score raw source fraction 0.4102; unfair-internet flagged rate 0.008 source fraction (0.8%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Opus 4.8"; harness "claude-code"; source effort "low"; Score raw source fraction 0.3514; unfair-internet flagged rate 0.001 source fraction (0.1%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Opus 4.8"; harness "claude-code"; source effort "max"; Score raw source fraction 0.465; unfair-internet flagged rate 0.006 source fraction (0.6%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Opus 4.8"; harness "claude-code"; source effort "medium"; Score raw source fraction 0.4046; unfair-internet flagged rate 0.005 source fraction (0.5%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Opus 4.8"; harness "claude-code"; source effort "xhigh"; Score raw source fraction 0.4549; unfair-internet flagged rate 0.005 source fraction (0.5%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Opus 5"; harness "claude-code"; source effort "high"; Score raw source fraction 0.4799; unfair-internet flagged rate 0.002 source fraction (0.2%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Opus 5"; harness "claude-code"; source effort "low"; Score raw source fraction 0.4195; unfair-internet flagged rate 0.0021 source fraction (0.21%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Opus 5"; harness "claude-code"; source effort "max"; Score raw source fraction 0.4804; unfair-internet flagged rate 0.0101 source fraction (1.01%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Opus 5"; harness "claude-code"; source effort "medium"; Score raw source fraction 0.5338; unfair-internet flagged rate 0.006 source fraction (0.6%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Opus 5"; harness "claude-code"; source effort "xhigh"; Score raw source fraction 0.4365; unfair-internet flagged rate 0.0144 source fraction (1.44%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Opus 5.5"; harness "claude-code"; source effort "high"; Score raw source fraction 0.5399; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Opus 5.5"; harness "claude-code"; source effort "low"; Score raw source fraction 0.473; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Opus 5.5"; harness "claude-code"; source effort "max"; Score raw source fraction 0.5443; unfair-internet flagged rate 0.008 source fraction (0.8%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Opus 5.5"; harness "claude-code"; source effort "medium"; Score raw source fraction 0.5464; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Opus 5.5"; harness "claude-code"; source effort "xhigh"; Score raw source fraction 0.5142; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Sonnet 4.6"; harness "claude-code"; source effort "high"; Score raw source fraction 0.2347; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Sonnet 4.6"; harness "claude-code"; source effort "low"; Score raw source fraction 0.1503; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Sonnet 4.6"; harness "claude-code"; source effort "max"; Score raw source fraction 0.2431; unfair-internet flagged rate 0.002 source fraction (0.2%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Sonnet 4.6"; harness "claude-code"; source effort "medium"; Score raw source fraction 0.211; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Sonnet 5"; harness "claude-code"; source effort "high"; Score raw source fraction 0.3935; unfair-internet flagged rate 0.002 source fraction (0.2%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Sonnet 5"; harness "claude-code"; source effort "low"; Score raw source fraction 0.2871; unfair-internet flagged rate 0.001 source fraction (0.1%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Sonnet 5"; harness "claude-code"; source effort "max"; Score raw source fraction 0.4236; unfair-internet flagged rate 0.003 source fraction (0.3%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Sonnet 5"; harness "claude-code"; source effort "medium"; Score raw source fraction 0.352; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Claude Sonnet 5"; harness "claude-code"; source effort "xhigh"; Score raw source fraction 0.4273; unfair-internet flagged rate 0.005 source fraction (0.5%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Composer 2.5"; harness "cursor-cli"; source effort "none"; Score raw source fraction 0.2564; unfair-internet flagged rate 0.024 source fraction (2.4%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "DeepSeek V4 Flash 0731"; harness "chisel"; source effort "high"; Score raw source fraction 0.188; unfair-internet flagged rate 0.2547 source fraction (25.47%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "DeepSeek V4 Pro"; harness "mini-swe-agent"; source effort "none"; Score raw source fraction 0.1764; unfair-internet flagged rate 0.01 source fraction (1%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "DeepSeek V4 Pro 0813"; harness "chisel"; source effort "high"; Score raw source fraction 0.2855; unfair-internet flagged rate 0.106 source fraction (10.6%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Gemini 3.6 Flash"; harness "chisel"; source effort "high"; Score raw source fraction 0.3369; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Gemini 3.6 Flash"; harness "chisel"; source effort "low"; Score raw source fraction 0.2282; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Gemini 3.6 Flash"; harness "chisel"; source effort "medium"; Score raw source fraction 0.3437; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Gemini 3.7 Flash"; harness "chisel"; source effort "high"; Score raw source fraction 0.4223; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Gemini 3.7 Flash"; harness "chisel"; source effort "low"; Score raw source fraction 0.3691; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Gemini 3.7 Flash"; harness "chisel"; source effort "medium"; Score raw source fraction 0.4359; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Gemini 3.8 Flash"; harness "chisel"; source effort "high"; Score raw source fraction 0.3796; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Gemini 3.8 Flash"; harness "chisel"; source effort "medium"; Score raw source fraction 0.4119; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GLM 5.2"; harness "mini-swe-agent"; source effort "none"; Score raw source fraction 0.245; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GLM 5.3"; harness "chisel"; source effort "max"; Score raw source fraction 0.4014; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GLM 5.3 Flash"; harness "chisel"; source effort "max"; Score raw source fraction 0.3183; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-5.4-mini"; harness "codex"; source effort "high"; Score raw source fraction 0.2308; unfair-internet flagged rate 0.001 source fraction (0.1%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-5.4-mini"; harness "codex"; source effort "low"; Score raw source fraction 0.0925; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-5.4-mini"; harness "codex"; source effort "medium"; Score raw source fraction 0.2078; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-5.4-mini"; harness "codex"; source effort "xhigh"; Score raw source fraction 0.2704; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-5.5"; harness "codex"; source effort "high"; Score raw source fraction 0.4056; unfair-internet flagged rate 0.003 source fraction (0.3%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-5.5"; harness "codex"; source effort "low"; Score raw source fraction 0.3055; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-5.5"; harness "codex"; source effort "medium"; Score raw source fraction 0.366; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-5.5"; harness "codex"; source effort "xhigh"; Score raw source fraction 0.4296; unfair-internet flagged rate 0.004 source fraction (0.4%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-5.6 Luna"; harness "codex"; source effort "high"; Score raw source fraction 0.3589; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-5.6 Luna"; harness "codex"; source effort "low"; Score raw source fraction 0.1538; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-5.6 Luna"; harness "codex"; source effort "max"; Score raw source fraction 0.3981; unfair-internet flagged rate 0.001 source fraction (0.1%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-5.6 Luna"; harness "codex"; source effort "medium"; Score raw source fraction 0.2574; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-5.6 Luna"; harness "codex"; source effort "xhigh"; Score raw source fraction 0.3891; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-5.6 Sol"; harness "codex"; source effort "high"; Score raw source fraction 0.4506; unfair-internet flagged rate 0.0013 source fraction (0.13%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-5.6 Sol"; harness "codex"; source effort "low"; Score raw source fraction 0.3544; unfair-internet flagged rate 0.001 source fraction (0.1%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-5.6 Sol"; harness "codex"; source effort "max"; Score raw source fraction 0.4749; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-5.6 Sol"; harness "codex"; source effort "medium"; Score raw source fraction 0.3993; unfair-internet flagged rate 0.002 source fraction (0.2%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-5.6 Sol"; harness "codex"; source effort "xhigh"; Score raw source fraction 0.4684; unfair-internet flagged rate 0.001 source fraction (0.1%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-5.6 Terra"; harness "codex"; source effort "high"; Score raw source fraction 0.3689; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-5.6 Terra"; harness "codex"; source effort "low"; Score raw source fraction 0.2411; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-5.6 Terra"; harness "codex"; source effort "max"; Score raw source fraction 0.4131; unfair-internet flagged rate 0.001 source fraction (0.1%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-5.6 Terra"; harness "codex"; source effort "medium"; Score raw source fraction 0.3388; unfair-internet flagged rate 0.001 source fraction (0.1%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-5.6 Terra"; harness "codex"; source effort "xhigh"; Score raw source fraction 0.3875; unfair-internet flagged rate 0.001 source fraction (0.1%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-6 Astra"; harness "codex"; source effort "high"; Score raw source fraction 0.5094; unfair-internet flagged rate not reported; attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-6 Astra"; harness "codex"; source effort "low"; Score raw source fraction 0.4527; unfair-internet flagged rate not reported; attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-6 Astra"; harness "codex"; source effort "max"; Score raw source fraction 0.5326; unfair-internet flagged rate not reported; attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-6 Astra"; harness "codex"; source effort "medium"; Score raw source fraction 0.4883; unfair-internet flagged rate not reported; attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-6 Astra"; harness "codex"; source effort "xhigh"; Score raw source fraction 0.5062; unfair-internet flagged rate not reported; attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-6 Luna"; harness "codex"; source effort "high"; Score raw source fraction 0.3726; unfair-internet flagged rate 0.002 source fraction (0.2%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-6 Luna"; harness "codex"; source effort "low"; Score raw source fraction 0.2566; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-6 Luna"; harness "codex"; source effort "max"; Score raw source fraction 0.4242; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-6 Luna"; harness "codex"; source effort "medium"; Score raw source fraction 0.3553; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-6 Luna"; harness "codex"; source effort "xhigh"; Score raw source fraction 0.371; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-6 Sol"; harness "codex"; source effort "high"; Score raw source fraction 0.477; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-6 Sol"; harness "codex"; source effort "low"; Score raw source fraction 0.3728; unfair-internet flagged rate 0.002 source fraction (0.2%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-6 Sol"; harness "codex"; source effort "max"; Score raw source fraction 0.4927; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-6 Sol"; harness "codex"; source effort "medium"; Score raw source fraction 0.4592; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "GPT-6 Sol"; harness "codex"; source effort "xhigh"; Score raw source fraction 0.4845; unfair-internet flagged rate 0.002 source fraction (0.2%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Grok 4.5"; harness "grok-build"; source effort "high"; Score raw source fraction 0.4244; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Grok 4.5"; harness "grok-build"; source effort "low"; Score raw source fraction 0.3786; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Grok 4.5"; harness "grok-build"; source effort "medium"; Score raw source fraction 0.4193; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Grok 4.6"; harness "grok-build"; source effort "high"; Score raw source fraction 0.4801; unfair-internet flagged rate 0.007 source fraction (0.7%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Grok 4.6"; harness "grok-build"; source effort "low"; Score raw source fraction 0.4011; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Grok 4.6"; harness "grok-build"; source effort "medium"; Score raw source fraction 0.4606; unfair-internet flagged rate 0.006 source fraction (0.6%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Grok 4.7"; harness "grok-build"; source effort "high"; Score raw source fraction 0.4759; unfair-internet flagged rate 0.013 source fraction (1.3%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Grok 4.7"; harness "grok-build"; source effort "xhigh"; Score raw source fraction 0.4559; unfair-internet flagged rate 0.008 source fraction (0.8%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Inkling"; harness "mini-swe-agent"; source effort "0.99"; Score raw source fraction 0.14; unfair-internet flagged rate 0.076 source fraction (7.6%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Kimi K2.7"; harness "mini-swe-agent"; source effort "none"; Score raw source fraction 0.3006; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Kimi K3"; harness "mini-swe-agent"; source effort "none"; Score raw source fraction 0.4417; unfair-internet flagged rate 0.0021 source fraction (0.21%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "MiniMax M3"; harness "mini-swe-agent"; source effort "none"; Score raw source fraction 0.147; unfair-internet flagged rate 0.015 source fraction (1.5%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Mistral 3.5 Medium"; harness "chisel"; source effort "none"; Score raw source fraction 0.08; unfair-internet flagged rate 0.006 source fraction (0.6%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Nemotron 3 Ultra"; harness "chisel"; source effort "none"; Score raw source fraction 0.1361; unfair-internet flagged rate 0.002 source fraction (0.2%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "Qwen 3.7 Plus"; harness "mini-swe-agent"; source effort "none"; Score raw source fraction 0.102; unfair-internet flagged rate 0.009 source fraction (0.9%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "SWE-1.6"; harness "chisel"; source effort "none"; Score raw source fraction 0.0939; unfair-internet flagged rate 0 source fraction (0%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "SWE-1.7"; harness "chisel"; source effort "none"; Score raw source fraction 0.4199; unfair-internet flagged rate 0.012 source fraction (1.2%); attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "SWE-2"; harness "devin"; source effort "high"; Score raw source fraction 0.472; unfair-internet flagged rate not reported; attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "SWE-2"; harness "devin"; source effort "max"; Score raw source fraction 0.5; unfair-internet flagged rate not reported; attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Official Cognition FrontierCode 1.1 Main result over 100 tasks; exact source model label "SWE-2"; harness "devin"; source effort "medium"; Score raw source fraction 0.4309; unfair-internet flagged rate not reported; attempt count and fallback behavior are not reported. Scores are retrieved 2026-09-23. Anthropic's table gives Opus 5.5 adaptive Max; comparison-model efforts are not fully specified in this table. The GPT-6 Astra and GPT-5.6 Sol values are publisher-reported comparison cells. Official public FrontierCode 1.1 leaderboard values.
Method / source