Just 3 Weeks Later: Gemini 3.7 Flash Shows Major Gains in Coding and Document Processing
Ars Technica AI · rss · 2026-08-14
Google has rapidly introduced Gemini 3.7 Flash merely three weeks after the release of 3.6 Flash. Positioned as a "workhorse" model, it achieves significant improvements in coding and agentic performance, coupled with a lower introductory price.
- Coding Performance: The FrontierCode 1.1 test score jumped from 34.4% to 43.6%, DeepSWE v1.1 went from 49% to 65.3%, and the WebDev Arena score rose to 1,588.
- Complex Docs & Automation: The GDP.pdf benchmark, measuring complex document processing, increased from 22% to 34%. AutomationBench, evaluating common business workflows, doubled its score from 17% to 30.4%.
More from Models
- Devs say LLM coding mistakes are a different class: overcomplication, inferred limits, worse at higher reasoning — Mu5_ · 2026-10-16
- ChatGPT UI screenshot reveals Wallet coming soon plus Email, Calendar, Health and Finances integrations — jxnlco · 2026-10-16
- Why the next OpenAI and Anthropic models may still disappoint: no native audio or video input — imjustnewatai · 2026-10-16
- Thousands Hit by ChatGPT Standard Voice Mode Regression: Wrong Voice, Interruptions — shaunralston · 2026-10-16
- Gemini Argon 4 Quota Exhausted, Reset Countdown Shows Nearly 7 Days — ssh4net · 2026-10-16
- AINews: Meta ships 816B Muse Spark Open; Goodfire argues neural nets are already clean — Latent Space · 2026-10-16