Gemini 3.8 Flash hits 73.7% on DeepSWE, nearly matching Opus 5 and GPT-5.6 Sol
cedric_chee · x · 2026-09-03
A user posted apparent DeepSWE coding benchmark scores for Gemini 3.8 Flash (high): 73.7% pass@1, essentially tied with frontier flagships — Opus 5 (max) at 74% pass@1, GPT-5.6 Sol (max) at 73% pass@1, and Fable 5.1 at 67.4% pass@5. If accurate, Google's lightweight Flash tier is now competitive with rivals' flagship models on coding tasks (unverified third-party figures).
More from Models
- Mocked benchmark shows DeepSeek's document chart grounding badly misaligned — burkov · 2026-10-18
- Dev complains Claude's habit of saying "not done" is his biggest pain point — jasonkneen · 2026-10-18
- Open-weight TYGER speech decision model makes 80ms turn-taking calls for full-duplex voice agents — solyarisoftware · 2026-10-18
- Paying Google AI subscriber blocked: custom-domain Workspace accounts can't access AI Studio — ChrisRackauckas · 2026-10-18
- Running Ling 3.1 Flash (560B MoE, 25B Active) Across 4 DGX Sparks via llama.cpp RPC — BerryBlush_Rogers691 · 2026-10-18
- Users press OpenAI on long-delayed Adult Mode a year after it was promised — ___Patrice___ · 2026-10-18