Frontier Log

Google

Gemini 2.5 Flash-Lite

Announced on blog.google

Benchmark results

BenchmarkScoreSource check
Humanity's Last ExamNo tools · thinking; pass@16.9%
CheckedA person checked this number against its source on 26 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
non-thinking; pass@15.1%
CheckedA person checked this number against its source on 26 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
SWE-bench Verifiednon-thinking; single attempt31.6%
CheckedA person checked this number against its source on 26 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
thinking; single attempt27.6%
CheckedA person checked this number against its source on 26 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
non-thinking; multiple attempts: Google's scaffold draws several trajectories and the model picks one · parallel test-time compute42.6%
CheckedA person checked this number against its source on 26 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
thinking; multiple attempts: Google's scaffold draws several trajectories and the model picks one · parallel test-time compute44.9%
CheckedA person checked this number against its source on 26 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
GPQA Diamondthinking; pass@166.7%
CheckedA person checked this number against its source on 26 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
non-thinking; pass@164.6%
CheckedA person checked this number against its source on 26 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
AIME 2025No tools · thinking; pass@163.1%
CheckedA person checked this number against its source on 26 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
non-thinking; pass@149.8%
CheckedA person checked this number against its source on 26 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
LiveCodeBenchthinking; pass@1; problems from 1/1/2025 to 5/1/202534.3%
CheckedA person checked this number against its source on 26 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
non-thinking; pass@1; problems from 1/1/2025 to 5/1/202533.7%
CheckedA person checked this number against its source on 26 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
Sources

Reported by Google

Related events

All models