Claude 3.7 Sonnet vs Claude 4 Opus Thinking
Change modelsClaude 3.7 SonnetAnthropic
Claude 4 Opus ThinkingAnthropic
IndexAll categories, out of 100.
37.7 ±4.0
47.0 ±7.6 best
GapIndex points below the top.
−9.3
Top
CoverageIndex weight that has results.
95% best
75%
SpeedOutput tokens per second.
—
—
EloLMArena rating and rank.
1299
—
ReleasedMonth of release.
2025-02
2025-05
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.
200K
200K
AgenticMulti-step tasks with tools.
40.7
51.5 best
CodingCode writing and repair.
42.1
—
ReasoningLogic problems and puzzles.
39.8
54.8 best
MultimodalTasks with images and text.
40.3
44.9 best
KnowledgeFacts and expert knowledge.
40.0
47.1 best
MultilingualTasks in many languages.
—
—
InstructionTasks with strict prompt rules.
38.1
39.2 best
MathMath problems.
30.0
43.9 best
Benchmarks
8 sharedOnly Claude 3.7 Sonnet: 34Only Claude 4 Opus Thinking: 3
Claude 3.7 Sonnet
Claude 4 Opus Thinking