GPT-5.6 gains on a detection leaderboard after images are downscaled to 2048×2048
bdsqlsz · x · 2026-07-24
After downscaling input images to a maximum of 2048×2048, GPT-5.6 improved on a detection leaderboard: GPT-5.6 Sol rose by 4.0% to 68.2% and reached #2, GPT-5.6 Terra rose by 8.3% to 60.7% and reached #4, and GPT-5.6 Luna rose by 7.8% to 59.9% and reached #6.
The image shows a before/after comparison of the leaderboard, with the three GPT-5.6 variants moving up after the image resize change.
More from Models
- Benchmark scores drop from 89% to 19% on new evals — how benchmaxxing breaks leaderboard trust — airesearch12 · 2026-09-11
- ChatGPT tells user their question is too hard and to 'accept dumber answers' — phido3000 · 2026-09-11
- Developer Building a Unified Leaderboard of All Model Benchmark Scores — airesearch12 · 2026-09-11
- Rumor claims Kimi faked performance by serving Claude; DeepSeek new model surprises in evals — realsohamparekh · 2026-09-11
- GPT-5.6 writes well but is instantly forgettable, user complains — BasedRaddka · 2026-09-11
- Opus Refuses Protein Research Codebase Over 'Safety' Concerns, Dev Considers Rolling His Own — josephdviviano · 2026-09-11