Gemini 3.8 Flash rolls out scoring 71% on DeepSWE at half-price promo until 2027
testingcatalog · x · 2026-09-02
Per testingcatalog, Google is rolling out Gemini 3.8 Flash across the Gemini app, Google AI Studio, and APIs.
Key points:
- Performance: scores 71% on DeepSWE 1.1, close to Claude Opus 5's 74%, at a much lower price
- Pricing: input $0.75/M tokens through Dec 31, 2026, then $1.50; output (incl. thinking tokens) $3.75/M, then $7.50
Essentially a limited-time half-price window before prices double in 2027.
Related event: Gemini 3.8 Flash goes GA, third Flash upgrade in six weeks(19 posts)→
More from Models
- Agent Arena launches leaderboard ranking 57 models across 2.18M agentic sessions — arena · 2026-09-03
- Gemini 3.8 Flash launches at same price and speed, called best general-purpose API model — theREALmarvin · 2026-09-02
- Early hands-on: Gemini 3.8 Flash impresses for coding inside Antigravity — fhinkel · 2026-09-02
- Google launches Gemini 3.8 Flash and a new 3.8 Flash Cyber variant — Jame92 · 2026-09-02
- Viral claim: an OpenAI model escaped its sandbox, hit Hugging Face; training reportedly paused — thetripathi58 · 2026-09-02
- 3.8 Flash Cyber launches as a low-cost cyber-specialized model — melvinjohnsonp · 2026-09-02