Claude Opus 4.8 vs Step 5 Preview

Change models
IndexAll categories, out of 100.
66.6 ±2.0
67.4 ±3.7 best
GapIndex points below the top.
−0.8 · no clear gap
Top
CoverageIndex weight that has results.
100% best
90%
SpeedOutput tokens per second.
76/s best
26/s
EloLMArena rating and rank.
1454 best
1444
ReleasedMonth of release.
2026-05
2026-09
Input / 1MUS dollars per 1M input tokens.
$5.00
$1.00 best
Output / 1MUS dollars per 1M output tokens.
$25.00
$2.70 best
Batch output / 1MBatch price per 1M output tokens.
$12.50
—
1K chats2K tokens in, 500 out, each.
$22.50
$3.35 best
1K documents20K tokens in, 1K out, each.
$125.00
$22.70 best
1K agent tasks200K tokens in, 20K out, each.
$1,500.00
$254.00 best
ContextMaximum tokens in one request.
1M
1M
AgenticMulti-step tasks with tools.
65.2
67.7 best
CodingCode writing and repair.
67.3
70.4 best
ReasoningLogic problems and puzzles.
66.6 best
66.2
MultimodalTasks with images and text.
66.6 best
61.4
KnowledgeFacts and expert knowledge.
65.5 best
63.0
MultilingualTasks in many languages.
59.2
—
InstructionTasks with strict prompt rules.
58.8
—
MathMath problems.
69.9 best
61.1

Benchmarks

30 shared
Only Claude Opus 4.8: 74Only Step 5 Preview: 32
Claude Opus 4.8
Step 5 Preview
48.7% best15 / 216
46.5%23 / 216
77.7%73 / 211
88.3% best2 / 211
20.9%25 / 207
20.9%25 / 207
48.8% best29 / 170
36.4%67 / 170
47.7%32 / 132
54.3% best24 / 132
82.7% best29 / 130
70.4%55 / 130
54.4%33 / 113
58.9% best13 / 113
43.8% best28 / 98
40.4%40 / 98
30.0% best37 / 89
27.3%42 / 89
74.6%47 / 89
85.0% best20 / 89
93.6% best10 / 87
93.5%12 / 87
66.0% best63 / 80
63.5%69 / 80
89.2% best19 / 80
83.8%44 / 80
53.9% best29 / 79
50.7%42 / 79
47.3% best16 / 77
39.0%31 / 77
57.9% best9 / 77
46.5%28 / 77
EMBFinance
69.4% best15 / 75
57.1%41 / 75
23.2%36 / 68
31.8% best28 / 68
BrowseCompAgentic
84.3%20 / 63
88.7% best12 / 63
MCP AtlasAgentic
82.2%13 / 53
85.6% best4 / 53