PHP
GOAL
Domain ladder: real PHP work.
WHAT IT TESTS
PHP fixes and features across a graded ladder.
HOW IT GRADES
Graded by phpunit.
Standings — gauntlet benches solved
| Model | Solved | |
|---|---|---|
| qwen3.6-27b | 10/10 | |
| qwen3-next-80b | 5/5 | |
| gpt-oss-120b | 5/5 | |
| gpt-oss-20b | 5/5 | |
| granite-4.1-30b | 5/5 | |
| llama-3.3-70b | 5/5 | |
| phi-4 | 5/5 | |
| qwen3-235b-a22b-q4-cluster | 5/5 | |
| qwen3.5-4b | 5/5 | |
| qwen3.6-35b-a3b | 5/5 | |
| qwen3.6-35b-a3b-or | 5/5 | |
| qwen3.6-35b-bf16 | 5/5 | |
| claude-sonnet-4.6 | cloud | 5/5 |
| gemma-4-26b-a4b-it | 5/5 | |
| mistral-small-3.1-24b-instruct-2503 | 4/5 | |
| deepseek-r1-32b | 4/5 | |
| devstral-small-2-2512 | 4/5 | |
| qwen3.5-122b-a10b-cluster | 4/5 | |
| granite-4.1-8b | 4/5 | |
| gemini-2.5-flash-lite | cloud | 3/5 |
| qwen3-coder-30b | 2/5 | |
| qwen3.5-2b | 2/5 | |
| phi-4-reasoning-plus | 1/5 | |
| qwen3.5-0.8b | 0/5 | |
| deepseek-r1-distill-qwen-32b | 0/5 |
Benches — green = in gauntlet
ladder.php.group_totalsladder.php.parse_amountladder.php.parse_kvladder.php.slugifyladder.php.template_render