Claude Sonnet 4 Thinking vs Gemini 2.5 Flash (Preview)
Change modelsClaude Sonnet 4 ThinkingAnthropic
IndexAll categories, out of 100.
46.8 ±8.2 best
46.5 ±6.7
GapIndex points below the top.
Top
−0.3 · no clear gap
CoverageIndex weight that has results.
75%
90% best
SpeedOutput tokens per second.
—
—
EloLMArena rating and rank.
—
—
ReleasedMonth of release.
—
—
Input / 1MUS dollars per 1M input tokens.
—
—
Output / 1MUS dollars per 1M output tokens.
—
—
1K chats2K tokens in, 500 out, each.
—
—
1K documents20K tokens in, 1K out, each.
—
—
1K agent tasks200K tokens in, 20K out, each.
—
—
ContextMaximum tokens in one request.
—
—
AgenticMulti-step tasks with tools.
50.7 best
40.8
CodingCode writing and repair.
41.7
51.5 best
ReasoningLogic problems and puzzles.
—
41.2
MultimodalTasks with images and text.
52.6
58.5 best
KnowledgeFacts and expert knowledge.
47.9
50.8 best
MultilingualTasks in many languages.
—
—
InstructionTasks with strict prompt rules.
—
—
MathMath problems.
45.9 best
37.2
Benchmarks
14 sharedOnly Claude Sonnet 4 Thinking: 2Only Gemini 2.5 Flash (Preview): 4
Claude Sonnet 4 Thinking
Gemini 2.5 Flash (Preview)