Google: Gemini 3.1 Flash Lite

Community sentiment

Negative (-16) — 3 positive, 6 negative, 3 neutral across 12 community samples from X and the web, as of 2026-10-22.

Public discussion of Gemini 3.1 Flash Lite is mixed: users praise its benchmark accuracy and cost/performance in comparisons and report that the API operates as expected. Criticisms center on reliability and formatting bugs (e.g., returning "null" as text), slower real-world throughput versus some alternatives, and an announced deprecation timeline.

• “7/N On BrowseComp-Plus, Librarian Agents reaches 75% accuracy with a Qwen3.5-4B answerer and 77% with Gemini 3.1 Flash Lite, versus 44% and 42% for the strongest baselines. https://t.co/PCj4AT4H3X” — X
• “Gemini 3.1 Flash-Lite vs Gemini 3.5 Flash-Lite: Comparison - BenchLM — 3 days ago · Flash-Lite has the higher public point estimate, 49.59 versus 46.98. Repository review: $0.017 vs $0.0225. Cache-heavy agent loop: $0.025 vs $0.037.” — the web
• “a required string has to come back as a string, for example: gemini 3.1 flash-lite wrote "null" as text on 5 of 5 runs, which lands in ur database as a real PO number to be fair, make the field nullable and all 4 return null. it's fixable, but it's one more thing to remember on e” — X
• “I love Mercury AI models because of the speed of response 1000 tokens/s 🤯 3x faster than my current main model - Gemini 3.1 Flash Lite Definitely worth testing Pricing: $0.20 in, $0.70 out https://t.co/Cgm5Kat57t” — X

RuntimeWire coverage

• Meta ships Muse Image API at $0.01 per image
• Google launches cheaper Gemini 3.6 Flash and gates cyber model access