← Harnesses
baseline
legacymaincurrentThe model-leaderboard harness — light, single-attempt scaffolding.
why it exists: The control line: minimal help so the leaderboard reflects the model, not the harness.
Models
25
have run it
Solved
808
gauntlet benches
Models on this harness (gauntlet solved / attempted)
| Model | Solved | By domain |
|---|---|---|
| gemma-4-26b-a4b-it | 43/43 | Agentic 6/6Coding 6/6C# 15/15Gate 11/11PHP 5/5 |
| qwen3.6-35b-a3b | 43/43 | Agentic 6/6Coding 6/6C# 15/15Gate 11/11PHP 5/5 |
| gpt-oss-20b | 42/43 | Agentic 6/6Coding 6/6C# 14/15Gate 11/11PHP 5/5 |
| qwen3-next-80b | 42/43 | Agentic 6/6Coding 6/6C# 14/15Gate 11/11PHP 5/5 |
| claude-sonnet-4.6 | 41/43 | Agentic 6/6Coding 6/6C# 13/15Gate 11/11PHP 5/5 |
| gpt-oss-120b | 41/43 | Agentic 5/6Coding 5/6C# 15/15Gate 11/11PHP 5/5 |
| qwen3.6-27b | 41/43 | Agentic 6/6Coding 6/6C# 13/15Gate 11/11PHP 5/5 |
| qwen3.6-35b-a3b-or | 41/43 | Agentic 6/6Coding 6/6C# 13/15Gate 11/11PHP 5/5 |
| qwen3.6-35b-bf16 | 41/43 | Agentic 6/6Coding 5/6C# 14/15Gate 11/11PHP 5/5 |
| granite-4.1-30b | 40/43 | Agentic 6/6Coding 6/6C# 12/15Gate 11/11PHP 5/5 |
| mistral-small-3.1-24b-instruct-2503 | 38/43 | Agentic 6/6Coding 5/6C# 12/15Gate 11/11PHP 4/5 |
| qwen3.5-122b-a10b-cluster | 38/43 | Agentic 6/6Coding 5/6C# 12/15Gate 11/11PHP 4/5 |
| devstral-small-2-2512 | 37/43 | Agentic 6/6Coding 5/6C# 11/15Gate 11/11PHP 4/5 |
| llama-3.3-70b | 37/41 | Agentic 6/6Coding 6/6C# 11/13Gate 9/11PHP 5/5 |
| qwen3.5-4b | 36/43 | Agentic 6/6Coding 3/6C# 11/15Gate 11/11PHP 5/5 |
| granite-4.1-8b | 35/43 | Agentic 6/6Coding 4/6C# 10/15Gate 11/11PHP 4/5 |
| qwen3-coder-30b | 35/43 | Agentic 6/6Coding 5/6C# 12/15Gate 10/11PHP 2/5 |
| gemini-2.5-flash-lite | 32/43 | Agentic 5/6Coding 4/6C# 9/15Gate 11/11PHP 3/5 |
| qwen3.5-2b | 25/43 | Agentic 6/6Coding 2/6C# 6/15Gate 9/11PHP 2/5 |
| deepseek-r1-distill-qwen-32b | 24/43 | Agentic 1/6Coding 6/6C# 10/15Gate 7/11PHP 0/5 |
| qwen3-235b-a22b-q4-cluster | 17/20 | C# 12/15PHP 5/5 |
| phi-4-reasoning-plus | 16/43 | Agentic 2/6Coding 4/6C# 5/15Gate 4/11PHP 1/5 |
| qwen3.5-0.8b | 14/43 | Agentic 5/6Coding 0/6C# 0/15Gate 9/11PHP 0/5 |
| phi-4 | 5/5 | PHP 5/5 |
| deepseek-r1-32b | 4/5 | PHP 4/5 |