← Harnesses
mini-blind
—external-calibrationmini-swe-agent, blind variant — external reference run.
why it exists: Calibration against a known minimal agent.
Models
11
have run it
Solved
0
gauntlet benches
Models on this harness (gauntlet solved / attempted)
| Model | Solved | By domain |
|---|---|---|
| deepseek-r1-32b | 0/0 | |
| devstral-small-2-2512 | 0/0 | |
| gemma-4-26b-a4b-it | 0/0 | |
| gpt-oss-20b | 0/0 | |
| granite-4.1-8b | 0/0 | |
| llama-3.3-70b | 0/0 | |
| mistral-small-3.1-24b-instruct-2503 | 0/0 | |
| phi-4 | 0/0 | |
| qwen3.6-35b-a3b | 0/0 | |
| qwen3-coder-30b | 0/0 | |
| qwen3-next-80b | 0/0 |