Inference
EN

Ling 3.1 Flash vs Step 5 Preview

Change models
Ling 3.1 FlashInclusionAI
Step 5 PreviewStepFun
IndexAll categories, out of 100.
67.2 ±7.2
68.4 ±3.7 best
GapIndex points below the top.
−1.2 · no clear gap
Top
CoverageIndex weight that has results.
70%
90% best
SpeedOutput tokens per second.
81/s
105/s best
EloLMArena rating and rank.
—
1443
ReleasedMonth of release.
2026-09
2026-09
Input / 1MUS dollars per 1M input tokens.
$0.045 best
$1.00
Output / 1MUS dollars per 1M output tokens.
$0.135 best
$2.70
1K chats2K tokens in, 500 out, each.
$0.158 best
$3.35
1K documents20K tokens in, 1K out, each.
$1.04 best
$22.70
1K agent tasks200K tokens in, 20K out, each.
$11.70 best
$254.00
ContextMaximum tokens in one request.
1M
1M
AgenticMulti-step tasks with tools.
66.6
68.8 best
CodingCode writing and repair.
66.9
71.4 best
ReasoningLogic problems and puzzles.
68.0 best
67.4
MultimodalTasks with images and text.
—
62.2
KnowledgeFacts and expert knowledge.
70.4 best
63.5
MultilingualTasks in many languages.
—
—
InstructionTasks with strict prompt rules.
—
—
MathMath problems.
—
61.7

Benchmarks

13 shared
Only Ling 3.1 Flash: 5Only Step 5 Preview: 49
Ling 3.1 Flash
Step 5 Preview
Artificial Analysis Humanity's Last ExamKnowledge
39.4%50 / 222
46.5% best23 / 222
Artificial Analysis Long Context ReasoningReasoning
83.0%18 / 217
88.3% best2 / 217
Critical Physics TasksReasoning
18.0%35 / 214
20.9% best26 / 214
GDPval-AA normalizedAgentic
55.9% best16 / 133
54.3%24 / 133
Artificial Analysis SciCodeCoding
54.1%36 / 114
58.9% best13 / 114
Terminal-Bench 2.1 · BenchLMAgentic
81.2%31 / 86
85.0% best20 / 86
Humanity's Last ExamKnowledge
37.4%37 / 77
46.5% best26 / 77
BrowseCompAgentic
91.7% best5 / 63
88.7%12 / 63
DeepSWEAgentic
59.7%42 / 59
67.7% best27 / 59
MCP AtlasAgentic
83.0%11 / 52
85.6% best4 / 52
AutomationBenchAgentic
52.5% best8 / 32
44.0%17 / 32
CyberGymAgentic
87.9% best4 / 29
84.7%8 / 29
Data Research and Analysis with Complex OperationsAgentic
85.5%
83.3%