Inference
EN

Gemini 4 Argon vs Step 5 Preview

Change models
Gemini 4 ArgonGoogle
Step 5 PreviewStepFun
IndexAll categories, out of 100.
78.5 ±4.3 best
68.4 ±3.7
GapIndex points below the top.
Top
−10.1
CoverageIndex weight that has results.
80%
90% best
SpeedOutput tokens per second.
28/s
105/s best
EloLMArena rating and rank.
1530 best
1443
ReleasedMonth of release.
2026-09
2026-09
Input / 1MUS dollars per 1M input tokens.
$2.00
$1.00 best
Output / 1MUS dollars per 1M output tokens.
$10.00
$2.70 best
Batch output / 1MBatch price per 1M output tokens.
$5.00
—
1K chats2K tokens in, 500 out, each.
$9.00
$3.35 best
1K documents20K tokens in, 1K out, each.
$50.00
$22.70 best
1K agent tasks200K tokens in, 20K out, each.
$600.00
$254.00 best
ContextMaximum tokens in one request.
1.05M best
1M
AgenticMulti-step tasks with tools.
77.3 best
68.8
CodingCode writing and repair.
76.7 best
71.4
ReasoningLogic problems and puzzles.
79.8 best
67.4
MultimodalTasks with images and text.
—
62.2
KnowledgeFacts and expert knowledge.
76.5 best
63.5
MultilingualTasks in many languages.
—
—
InstructionTasks with strict prompt rules.
—
—
MathMath problems.
75.9 best
61.7

Benchmarks

35 shared
Only Gemini 4 Argon: 51Only Step 5 Preview: 27
Gemini 4 Argon
Step 5 Preview
Artificial Analysis Humanity's Last ExamKnowledge
57.1% best4 / 222
46.5%23 / 222
Artificial Analysis Long Context ReasoningReasoning
79.7%45 / 217
88.3% best2 / 217
Critical Physics TasksReasoning
27.1% best17 / 214
20.9%26 / 214
Artificial Analysis Omniscience AccuracyKnowledge
48.3% best26 / 164
36.4%64 / 164
GDPval-AA normalizedAgentic
56.2% best14 / 133
54.3%24 / 133
Vibe Code Bench v1.1Coding
91.9% best3 / 123
70.4%48 / 123
Artificial Analysis SciCodeCoding
61.8% best5 / 114
58.9%13 / 114
Legal Research BenchLegal
54.8% best5 / 90
40.4%35 / 90
Tax Agent BenchFinance
49.2% best4 / 82
27.3%39 / 82
LiveBench Data AnalysisReasoning
85.2% best1 / 80
63.5%69 / 80
LiveBench ReasoningReasoning
95.1% best1 / 80
83.8%44 / 80
Finance Agent (v2)Finance
65.4% best1 / 79
50.7%42 / 79
Code MigrationCoding
68.2% best2 / 77
39.0%31 / 77
EMBFinance
75.2% best4 / 75
57.1%41 / 75
Terminal-Bench 4.0Agentic
57.6% best8 / 61
31.8%23 / 61
DeepSWEAgentic
77.9% best3 / 59
67.7%27 / 59
Agent Arena steerabilityAgentic
13.2 best9 / 54
−0.930 / 54
Agent Arena task outcomeAgentic
13.2 best7 / 54
−1.535 / 54
ProofBench v1.1Math
99.0% best6 / 54
42.0%37 / 54
LiveBench Coding 2.0Coding
89.7% best15 / 52
85.6%22 / 52