Inference
EN

Claude Opus 4.7 vs Gemini 3.1 Pro

Change models
Claude Opus 4.7Anthropic
Gemini 3.1 ProGoogle
IndexAll categories, out of 100.
62.0 ±3.0
62.3 ±3.1 best
GapIndex points below the top.
−0.3 · no clear gap
Top
CoverageIndex weight that has results.
95%
95%
SpeedOutput tokens per second.
47/s
151/s best
EloLMArena rating and rank.
1483
—
ReleasedMonth of release.
2026-04
2026-02
Input / 1MUS dollars per 1M input tokens.
$5.00
$2.00 best
Output / 1MUS dollars per 1M output tokens.
$25.00
$12.00 best
Batch output / 1MBatch price per 1M output tokens.
$12.50
—
1K chats2K tokens in, 500 out, each.
$22.50
$10.00 best
1K documents20K tokens in, 1K out, each.
$125.00
$52.00 best
1K agent tasks200K tokens in, 20K out, each.
$1,500.00
$640.00 best
ContextMaximum tokens in one request.
1M
1M
AgenticMulti-step tasks with tools.
60.6 best
58.4
CodingCode writing and repair.
64.5 best
58.0
ReasoningLogic problems and puzzles.
61.1
65.1 best
MultimodalTasks with images and text.
61.3
63.5 best
KnowledgeFacts and expert knowledge.
61.8
69.4 best
MultilingualTasks in many languages.
—
—
InstructionTasks with strict prompt rules.
45.7
61.1 best
MathMath problems.
64.3 best
61.7

Benchmarks

28 shared
Only Claude Opus 4.7: 37Only Gemini 3.1 Pro: 22
Claude Opus 4.7
Gemini 3.1 Pro
Artificial Analysis Humanity's Last ExamKnowledge
33.3%70 / 222
47.0% best21 / 222
Artificial Analysis Long Context ReasoningReasoning
75.7%92 / 217
82.0% best29 / 217
GPQA diamond · Epoch AIKnowledge
86.4%66 / 217
94.4% best9 / 217
Critical Physics TasksReasoning
5.1%83 / 214
17.7% best37 / 214
OTIS Mock AIME 2024-2025Math
86.7%69 / 200
95.6% best40 / 200
Artificial Analysis GPQA DiamondKnowledge
88.5%50 / 199
94.1% best5 / 199
τ²-Bench Tool-Agent-User EvaluationAgentic
74.0%78 / 149
95.6% best18 / 149
Artificial Analysis IFBenchInstruction
43.6%91 / 146
77.1% best10 / 146
Artificial Analysis MMMU-ProMultimodal
76.4%44 / 113
82.4% best14 / 113
SimpleQA VerifiedKnowledge
51.7%28 / 91
73.5% best6 / 91
Chess PuzzlesReasoning
41.0%34 / 90
51.0% best18 / 90
FrontierMath-Tiers-1-3-v2-PrivateMath
70.2% best29 / 90
59.6%42 / 90
Mystery Game PuzzlesReasoning
28.0%34 / 80
34.0% best25 / 80
FrontierMath-Tier-4-v2-PrivateMath
31.7% best32 / 72
26.8%39 / 72
FrontierMath-2025-02-28-PrivateMath
43.8% best6 / 69
36.9%13 / 69
FrontierMath-2025-02-28-PublicMath
90.0% best2 / 64
88.9%5 / 64
FrontierMath-Tier-4-2025-07-01-PrivateMath
22.9% best8 / 55
16.7%12 / 55
Gert Labs Composite Game BenchmarkAgentic
65.6% best3 / 52
56.9%14 / 52
MirrorCodeCoding
31.1% best30 / 39
8.9%38 / 39
ParseBench: document parsing for AI agents (LlamaIndex leaderboard)Multimodal
63.3%23 / 39
67.3% best17 / 39