Thinking Machines Inkling Launch - FORTRESS Benign: leaderboard
FORTRESS benign look-alike prompts (Scale AI) run by Thinking Machines: the rate at which a model still answers safe requests that resemble harmful ones, so higher means less over-refusal. Thinking Machines' Inkling launch post (July 2026), Safety rows of the full comparison table; percent, higher is better; Inkling models at effort 0.99.
Metric: Benign-Request Answer Rate (%). Source: thinkingmachines.ai. Saturation forecast: Not forecast. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | DeepSeek V4 Pro | 98.5 | #112 |
| 2 | Kimi K2.5 | 98.3 | #152 |
| 3 | GPT-5.6 Sol | 98.1 | #22 |
| 4 | Gemini 3.1 Pro (Preview) (High) | 98 | #64 (Gemini 3.1 Pro (Preview)) |
| 5 | Kimi K2.6 | 97.2 | #113 |
| 6 | Nemotron 3 Ultra | 90.5 | #245 |
| 7 | GLM-5.2 | 90 | #80 |
| 8 | Claude Fable 5 (Max) | 55.1 | #20 (Claude Fable 5) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=thinking-machines-inkling-launch-fortress-benign · How It Works · Data refreshed daily, snapshot 2026-11-05.