Step 3.7 Flash vs Step 5 Preview

Change models
IndexAll categories, out of 100.
55.3 ±5.0
68.4 ±3.7 best
GapIndex points below the top.
−13.1
Top
CoverageIndex weight that has results.
85%
90% best
SpeedOutput tokens per second.
112/s best
105/s
EloLMArena rating and rank.
—
1443
ReleasedMonth of release.
2026-05
2026-09
Input / 1MUS dollars per 1M input tokens.
$0.20 best
$1.00
Output / 1MUS dollars per 1M output tokens.
$1.15 best
$2.70
1K chats2K tokens in, 500 out, each.
$0.975 best
$3.35
1K documents20K tokens in, 1K out, each.
$5.15 best
$22.70
1K agent tasks200K tokens in, 20K out, each.
$63.00 best
$254.00
ContextMaximum tokens in one request.
262K
1M best
AgenticMulti-step tasks with tools.
59.6
68.8 best
CodingCode writing and repair.
54.9
71.4 best
ReasoningLogic problems and puzzles.
49.6
67.4 best
MultimodalTasks with images and text.
59.3
62.2 best
KnowledgeFacts and expert knowledge.
46.8
63.5 best
MultilingualTasks in many languages.
—
—
InstructionTasks with strict prompt rules.
51.9
—
MathMath problems.
—
61.7

Benchmarks

10 shared
Only Step 3.7 Flash: 11Only Step 5 Preview: 52
Step 3.7 Flash
Step 5 Preview
21.4%110 / 222
46.5% best23 / 222
73.7%100 / 217
88.3% best2 / 217
2.3%104 / 214
20.9% best26 / 214
25.8%101 / 164
36.4% best64 / 164
25.8%69 / 133
54.3% best24 / 133
43.9%78 / 114
58.9% best13 / 114
75.3%52 / 113
76.4% best44 / 113
59.5%65 / 86
85.0% best20 / 86
BrowseCompAgentic
75.8%37 / 63
88.7% best12 / 63
14.8%23 / 30
38.0% best5 / 30