Terminal-Bench Hard ↗53.03%
GPT-5.3 Codex (Xhigh) · Terminal-Bench Hard ↗
OpenAI / MODEL EVIDENCE
GPT-5.3 Codex
12 benchmarks · 24 observations · 16 sources · released 2026-02-05
Scores at a glance
12 benchmarks · large dot: GPT-5.3 Codex; small dots: closest models where reported
SWE-bench Verified ↗74.79%
GPT-5.3 Codex (high) · SWE-Bench verified · Best score (across scorers) · % ↗
1 more reported result
78 % · GPT 5.3 Codex (Mini-SWE-agent) · SWE-bench Verified · Score · Vals AI · SWE-bench Verified ↗
Terminal-Bench 2.0 ↗78.43%
GPT-5.3 Codex · Terminal Bench · Accuracy mean · % ↗
9 more reported results
77.3% · GPT-5.3 Codex · Terminal Bench · Accuracy mean · % · Epoch AI · Terminal Bench ↗
75.84% · GPT-5.3 Codex · Terminal Bench · Accuracy mean · % · Epoch AI · Terminal Bench ↗
75.73% · GPT-5.3 Codex · Terminal Bench · Accuracy mean · % · Epoch AI · Terminal Bench ↗
75.06% · GPT-5.3 Codex · Terminal Bench · Accuracy mean · % · Epoch AI · Terminal Bench ↗
74.61% · GPT-5.3 Codex · Terminal Bench · Accuracy mean · % · Epoch AI · Terminal Bench ↗
71.46% · GPT-5.3 Codex · Terminal Bench · Accuracy mean · % · Epoch AI · Terminal Bench ↗
69.14% · GPT-5.3 Codex · Terminal Bench · Accuracy mean · % · Epoch AI · Terminal Bench ↗
64.72% · GPT-5.3 Codex · Terminal Bench · Accuracy mean · % · Epoch AI · Terminal Bench ↗
70.3% · GPT-5.3 Codex · Terminal Bench · Accuracy mean · % · Epoch AI · Terminal Bench ↗
WeirdML (v2) ↗77.9%
GPT-5.3 Codex (xhigh) · WeirdML · Accuracy · % ↗
1 more reported result
79.3% · GPT-5.3 Codex · WeirdML · Accuracy · % · Epoch AI · WeirdML ↗
GPT-5.3 Codex · METR task performance & time horizons · Average task accuracy ↗
1 more reported result
349.53073 minutes · GPT-5.3 Codex · METR task performance & time horizons · 50% success task horizon · METR-Horizon-v1.1 · Epoch AI · METR Autonomy / Task-Horizon Suite · public result data ↗
Text Arena (Coding) ↗1369.86 Elo
GPT-5.3 Codex · WebDev Arena · Elo ↗
Epoch Capabilities Index (ECI) ↗157.09 ECI
GPT-5.3 Codex · General ECI · 26 October 2026 fit ↗
Vibe Code Bench ↗61.77 %
GPT 5.3 Codex (OpenHands) · Vibe Code Bench · original task set ↗
IOI v1 & IOI v2 ↗53.83 %
GPT 5.3 Codex · IOI · Accuracy ↗
LiveCodeBench ↗87.31 %
GPT 5.3 Codex · LiveCodeBench · Accuracy ↗
GPT 5.3 Codex · HiL-Bench (Human-in-Loop Benchmark) · Score ↗
Vending-Bench 2 ↗5940.124 USD
GPT-5.3-Codex · End-of-year balance ↗
Closest models by score
Similar positions on shared benchmark leaderboards. Smaller gaps are closer.
GPT-5.4 →OpenAI
4 of 6 shared benchmark comparisons
Claude Opus 4.6 →Anthropic
4 of 6 shared benchmark comparisons
GPT-5.6 Sol →OpenAI
4 of 6 shared benchmark comparisons
How similarity is measured
Mean absolute gap in percentile positions on shared benchmark settings; lower is closer. At least 60% overlap is required, with a minimum of three when available. Independent evaluations are used when there are at least three; otherwise published report comparisons are used. Repeated model observations within one setting are averaged before ranking. Writing-style traits and composite indices are excluded. This describes similarity, not an overall quality ranking.
Sources
Every plotted observation links to its evaluation. Reported configurations, benchmark versions and task subsets can differ; similarity describes shared scores, not training lift.
- Arena leaderboard — WebDev evaluation ↗
- Terminal-Bench Hard Benchmark | Artificial Analysis ↗
- METR Time Horizons ↗
- SWE-bench Verified ↗
- Terminal-Bench 2.0 ↗
- Vending Bench 2 — methodology and published results ↗
- WeirdML (v2) ↗
- AI benchmarks & capabilities ↗
- WeirdML ↗
- HiL-Bench (Human-in-Loop Benchmark) — evaluation methodology and results ↗
- Metr Autonomy Task Horizon Suite — methodology and published results ↗
- Terminal-Bench 2.0 leaderboard ↗
- IOI — evaluation methodology and results ↗
- LiveCodeBench — evaluation methodology and results ↗
- SWE-bench Verified — evaluation methodology and results ↗
- Vibe Code Bench — evaluation methodology and results ↗
Names used in source tables
GPT 5.3 Codex · GPT 5.3 Codex (Mini-SWE-agent) · GPT 5.3 Codex (OpenHands) · GPT-5.3 Codex · GPT-5.3 Codex (Xhigh) · GPT-5.3 Codex (high) · GPT-5.3 Codex (xhigh) · GPT-5.3-Codex