SWE-Bench Pro V2 Hard nearly saturated at 98% on day one
141_1337 · reddit · 2026-09-23
A Reddit user highlighted that SWE-Bench Pro V2 Hard was already nearly saturated at 98% by top models on its very first day, meaning the benchmark can no longer meaningfully differentiate frontier coding models and underscoring the accelerating pace of benchmark saturation.
Related event: SWE-Bench Pro V2 Hard Hits 98% on Day One, Instantly Saturated(2 posts)→
More from Models
- Robot build test: GPT-6 Astra stabilizes two-board structure, Claude Fable 5.2 falls short — qinzytech · 2026-11-12
- Zvi dissects Anthropic's latest system card, disputing the autonomy threshold call — Don't Worry About the Vase (Zvi) · 2026-11-12
- Unsloth ships Qwen3.8-27B Dynamic V3.1 GGUFs with 5-10% accuracy gains — danielhanchen · 2026-11-12
- antirez backs low-bit DeepSeek 4.1 Flash over Qwen 27B dense on Apple M5 Macs — antirez · 2026-11-12
- Gemini re-renders images as text when vision is removed, dropping only 6% WSR — zacharynado · 2026-11-12
- Bare-harness coding benchmark: Fable 5.2 best overall, Gemini 4 Carbon best value — Semantic_meaning · 2026-11-12