AI Studio API model id gemini-2.5-pro; default sampling settings. Technical report states Gemini scores are pass@1 and single-attempt unless specified. Aider Polyglot averages 3 trials; SWE-bench uses the internal agentic harness; Vibe-Eval uses Gemini as judge.
未知项
Thinking budget and any non-default settings for individual evaluations are not stated except where noted in each protocol.
工具
工具条件未记录
推理预算
未记录
框架
未记录
协议
Diamond subset; single attempt / pass@1.
条件
AI Studio API model gemini-2.5-pro; default sampling settings.
Gemini API model id gemini-3-pro-preview; default sampling settings. Gemini results are pass@1 and single-attempt unless specified; smaller benchmarks average multiple trials.
未知项
The evaluation table does not specify a thinking budget or inference-time compute setting.
工具
禁止工具
推理预算
未记录
框架
未记录
协议
No tools
条件
Scientific knowledge; Gemini API default sampling settings. 原材料协议说明:No tools.
Gemini API model id gemini-3.1-pro-preview; default sampling settings; evaluation table column specifies Thinking (High). All scores are pass@1 except where noted.
未知项
Exact thinking budget and sampling parameters are not listed.
工具
禁止工具
推理预算
high
框架
未记录
协议
No tools
条件
Gemini 3.1 Pro Thinking (High); Gemini API default sampling settings. 原材料协议说明:Diamond subset; no tools; self-computed by Google DeepMind.
Differences are calculated only for unique, active, exact values between direct predecessors when the unit, metric, and explicitly recorded configuration and conditions match.
· 1 score record for this release, including 1 exact value (may include historical records)
Gemini 2.5 Pro (GA) · GPQA Diamond scores and configurations
Metric and score
Configuration and evaluation conditions
Sources
Accuracy86.4%Reported 86.4% · Unit %
Published Unconfirmed
Gemini 2.5 Pro stable API
Settings
AI Studio API model id gemini-2.5-pro; default sampling settings. Technical report states Gemini scores are pass@1 and single-attempt unless specified. Aider Polyglot averages 3 trials; SWE-bench uses the internal agentic harness; Vibe-Eval uses Gemini as judge.
Unknowns
Thinking budget and any non-default settings for individual evaluations are not stated except where noted in each protocol.
Tools
Tool access not recorded
Reasoning budget
not recorded
Framework
not recorded
Protocol
Diamond subset; single attempt / pass@1.
Conditions
AI Studio API model gemini-2.5-pro; default sampling settings.
· 1 score record for this release, including 1 exact value (may include historical records)
Records for the direct predecessor do not provide matching conditions for a valid difference.
Gemini 3 Pro · GPQA Diamond scores and configurations
Metric and score
Configuration and evaluation conditions
Sources
Accuracy91.9%Reported 91.9% · Unit %
Published Unconfirmed
Gemini 3 Pro API default
Settings
Gemini API model id gemini-3-pro-preview; default sampling settings. Gemini results are pass@1 and single-attempt unless specified; smaller benchmarks average multiple trials.
Unknowns
The evaluation table does not specify a thinking budget or inference-time compute setting.
Tools
Tools forbidden
Reasoning budget
not recorded
Framework
not recorded
Protocol
No tools
Conditions
Scientific knowledge; Gemini API default sampling settings. 原材料协议说明:No tools.
· 1 score record for this release, including 1 exact value (may include historical records)
Records for the direct predecessor do not provide matching conditions for a valid difference.
Gemini 3.1 Pro · GPQA Diamond scores and configurations
Metric and score
Configuration and evaluation conditions
Sources
Accuracy94.3%Reported 94.3% · Unit %
Published Unconfirmed
Gemini 3.1 Pro Thinking (High)
Settings
Gemini API model id gemini-3.1-pro-preview; default sampling settings; evaluation table column specifies Thinking (High). All scores are pass@1 except where noted.
Unknowns
Exact thinking budget and sampling parameters are not listed.
Tools
Tools forbidden
Reasoning budget
high
Framework
not recorded
Protocol
No tools
Conditions
Gemini 3.1 Pro Thinking (High); Gemini API default sampling settings. 原材料协议说明:Diamond subset; no tools; self-computed by Google DeepMind.