StepFun: Step 3.5 Flash

Step 3.5 Flash is StepFun's most capable open-source foundation model. Built on a sparse Mixture of Experts (MoE) architecture, it selectively activates only 11B of its 196B parameters per token....

stepfun/step-3.5-flash

Step 3.5 Flash is StepFun's most capable open-source foundation model. Built on a sparse Mixture of Experts (MoE) architecture, it selectively activates only 11B of its 196B parameters per token....

Specifications & capabilities

Input
text
Output
text
Context window · tokens
262144
Published model ID
step-3.5-flash
Capabilities
promptCaching, reasoning, streaming, structuredOutputs, toolCalling

Evaluations

1–20 / 95

OpenCompass LLM

83.9%

source id
openCompass
leaderboard name
LiveCodeBenchV6
dimension
general
metric
LiveCodeBenchV6
rank
2
aggregation scope
status
published
direction
higherIsBetter
conditions
suite
OverallTable
catalog
open
snapshot
month
REALTIME
fileName
assets/research-rank/research-data.REALTIME.20260309.json
updateTime
2026-03-09
columnFileName
assets/research-rank/research-column.REALTIME.20260309.json
metric_kind
LiveCodeBenchV6
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
source_file
0-data.json
source_revision
117c5effa3066abf39f3e8f61109707b09f0cd95223246c07acd6289f4f7fcd2
metric_population
48
benchmark_evaluation
true
source model id
Step-3.5-Flash
Original result ↗

OpenCompass LLM

95.7%

source id
openCompass
leaderboard name
AIME2025
dimension
general
metric
AIME2025
rank
3
aggregation scope
status
published
direction
higherIsBetter
conditions
suite
OverallTable
catalog
open
snapshot
month
REALTIME
fileName
assets/research-rank/research-data.REALTIME.20260309.json
updateTime
2026-03-09
columnFileName
assets/research-rank/research-column.REALTIME.20260309.json
metric_kind
AIME2025
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
source_file
0-data.json
source_revision
117c5effa3066abf39f3e8f61109707b09f0cd95223246c07acd6289f4f7fcd2
metric_population
48
benchmark_evaluation
true
source model id
Step-3.5-Flash
Original result ↗

OpenCompass LLM

93.2%

source id
openCompass
leaderboard name
IFEval
dimension
general
metric
IFEval
rank
5
aggregation scope
status
published
direction
higherIsBetter
conditions
suite
OverallTable
catalog
open
snapshot
month
REALTIME
fileName
assets/research-rank/research-data.REALTIME.20260309.json
updateTime
2026-03-09
columnFileName
assets/research-rank/research-column.REALTIME.20260309.json
metric_kind
IFEval
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
source_file
0-data.json
source_revision
117c5effa3066abf39f3e8f61109707b09f0cd95223246c07acd6289f4f7fcd2
metric_population
48
benchmark_evaluation
true
source model id
Step-3.5-Flash
Original result ↗

OpenCompass LLM

76.9%

source id
openCompass
leaderboard name
Average
dimension
general
metric
Average
rank
8
aggregation scope
status
published
direction
higherIsBetter
conditions
suite
OverallTable
catalog
open
snapshot
month
REALTIME
fileName
assets/research-rank/research-data.REALTIME.20260309.json
updateTime
2026-03-09
columnFileName
assets/research-rank/research-column.REALTIME.20260309.json
metric_kind
Average
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
source_file
0-data.json
source_revision
117c5effa3066abf39f3e8f61109707b09f0cd95223246c07acd6289f4f7fcd2
metric_population
48
benchmark_evaluation
true
source model id
Step-3.5-Flash
Original result ↗

OpenCompass LLM

21.6%

source id
openCompass
leaderboard name
HLE
dimension
general
metric
HLE
rank
10
aggregation scope
status
published
direction
higherIsBetter
conditions
suite
OverallTable
catalog
open
snapshot
month
REALTIME
fileName
assets/research-rank/research-data.REALTIME.20260309.json
updateTime
2026-03-09
columnFileName
assets/research-rank/research-column.REALTIME.20260309.json
metric_kind
HLE
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
source_file
0-data.json
source_revision
117c5effa3066abf39f3e8f61109707b09f0cd95223246c07acd6289f4f7fcd2
metric_population
48
benchmark_evaluation
true
source model id
Step-3.5-Flash
Original result ↗

OpenCompass LLM

83.7%

source id
openCompass
leaderboard name
GPQA-Diamond
dimension
general
metric
GPQA-Diamond
rank
11
aggregation scope
status
published
direction
higherIsBetter
conditions
suite
OverallTable
catalog
open
snapshot
month
REALTIME
fileName
assets/research-rank/research-data.REALTIME.20260309.json
updateTime
2026-03-09
columnFileName
assets/research-rank/research-column.REALTIME.20260309.json
metric_kind
GPQA-Diamond
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
source_file
0-data.json
source_revision
117c5effa3066abf39f3e8f61109707b09f0cd95223246c07acd6289f4f7fcd2
metric_population
48
benchmark_evaluation
true
source model id
Step-3.5-Flash
Original result ↗

OpenCompass LLM

83.5%

source id
openCompass
leaderboard name
MMLU-Pro
dimension
general
metric
MMLU-Pro
rank
13
aggregation scope
status
published
direction
higherIsBetter
conditions
suite
OverallTable
catalog
open
snapshot
month
REALTIME
fileName
assets/research-rank/research-data.REALTIME.20260309.json
updateTime
2026-03-09
columnFileName
assets/research-rank/research-column.REALTIME.20260309.json
metric_kind
MMLU-Pro
rank_method
competition rank among reported evaluations in identical conditions, including unlinked observations
source_file
0-data.json
source_revision
117c5effa3066abf39f3e8f61109707b09f0cd95223246c07acd6289f4f7fcd2
metric_population
48
benchmark_evaluation
true
source model id
Step-3.5-Flash
Original result ↗

LMArena

1,388.722

source id
lmarena
leaderboard name
LMArena Text Factuality German
dimension
factuality
metric
Arena rating
rank
43
sample size
607
evaluated at
2026-11-09T00:00:00.000Z
source model id
Step-3.5-Flash
confidence lower
1366.7414331116472
confidence upper
1410.7034614362565
Original result ↗

LMArena

1,366.022

source id
lmarena
leaderboard name
LMArena Text Factuality Korean
dimension
factuality
metric
Arena rating
rank
47
sample size
592
evaluated at
2026-11-09T00:00:00.000Z
source model id
Step-3.5-Flash
confidence lower
1343.3581000331205
confidence upper
1388.6867174671618
Original result ↗

LMArena

1,369.369

source id
lmarena
leaderboard name
LMArena Text Factuality Japanese
dimension
factuality
metric
Arena rating
rank
49
sample size
427
evaluated at
2026-11-09T00:00:00.000Z
source model id
Step-3.5-Flash
confidence lower
1343.520880103713
confidence upper
1395.2172284611097
Original result ↗

LMArena

1,393.798

source id
lmarena
leaderboard name
LMArena Text Factuality Polish
dimension
factuality
metric
Arena rating
rank
63
sample size
889
evaluated at
2026-11-09T00:00:00.000Z
source model id
Step-3.5-Flash
confidence lower
1374.6628030836616
confidence upper
1412.9341044870775
Original result ↗

LMArena

1,448.755

source id
lmarena
leaderboard name
LMArena Text Factuality French
dimension
factuality
metric
Arena rating
rank
79
sample size
1600
evaluated at
2026-11-09T00:00:00.000Z
source model id
Step-3.5-Flash
confidence lower
1434.7512852933173
confidence upper
1462.7594203400913
Original result ↗

LMArena

1,415.173

source id
lmarena
leaderboard name
LMArena Text Factuality Spanish
dimension
factuality
metric
Arena rating
rank
87
sample size
1426
evaluated at
2026-11-09T00:00:00.000Z
source model id
Step-3.5-Flash
confidence lower
1399.5947982320517
confidence upper
1430.751006026669
Original result ↗

LMArena

1,428.492

source id
lmarena
leaderboard name
LMArena Text Factuality Industry Mathematical
dimension
factuality
metric
Arena rating
rank
95
sample size
2360
evaluated at
2026-11-09T00:00:00.000Z
source model id
Step-3.5-Flash
confidence lower
1417.202832038716
confidence upper
1439.7818848682914
Original result ↗

LMArena

1,465.015

source id
lmarena
leaderboard name
LMArena Text Factuality Chinese
dimension
factuality
metric
Arena rating
rank
99
sample size
2597
evaluated at
2026-11-09T00:00:00.000Z
source model id
Step-3.5-Flash
confidence lower
1454.3098897151717
confidence upper
1475.720261748414
Original result ↗

LMArena

1,420.598

source id
lmarena
leaderboard name
LMArena Text Factuality Math
dimension
factuality
metric
Arena rating
rank
102
sample size
2484
evaluated at
2026-11-09T00:00:00.000Z
source model id
Step-3.5-Flash
confidence lower
1409.2584758289443
confidence upper
1431.9373786088108
Original result ↗

LMArena

1,443.568

source id
lmarena
leaderboard name
LMArena Text Factuality Expert
dimension
factuality
metric
Arena rating
rank
110
sample size
3932
evaluated at
2026-11-09T00:00:00.000Z
source model id
Step-3.5-Flash
confidence lower
1434.0265934848387
confidence upper
1453.1096237168479
Original result ↗

LMArena

1,447.336

source id
lmarena
leaderboard name
LMArena Text Factuality Industry Medicine And Healthcare
dimension
factuality
metric
Arena rating
rank
111
sample size
3062
evaluated at
2026-11-09T00:00:00.000Z
source model id
Step-3.5-Flash
confidence lower
1436.930534904975
confidence upper
1457.7416465585006
Original result ↗

LMArena

1,429.472

source id
lmarena
leaderboard name
LMArena Text Factuality Industry Legal And Government
dimension
factuality
metric
Arena rating
rank
116
sample size
3464
evaluated at
2026-11-09T00:00:00.000Z
source model id
Step-3.5-Flash
confidence lower
1419.339306393036
confidence upper
1439.604925156986
Original result ↗

LMArena

1,418.527

source id
lmarena
leaderboard name
LMArena Text Spanish
dimension
general
metric
Arena rating
rank
119
sample size
2005
evaluated at
2026-11-09T00:00:00.000Z
source model id
Step-3.5-Flash
confidence lower
1404.6981405957927
confidence upper
1432.3557862312928
Original result ↗

Published prices

1–7 / 7

stepfun-ai/Step-3.5-Flash ↗

$0.10

Inference provider
siliconFlow
Platform
SiliconFlow
Unit
input_tokens_per_million
Currency
USD
Billing conditions
Billing label
Input tokens (per 1M)

stepfun-ai/Step-3.5-Flash ↗

$0.30

Inference provider
siliconFlow
Platform
SiliconFlow
Unit
output_tokens_per_million
Currency
USD
Billing conditions
Billing label
Output tokens (per 1M)

step-3.5-flash ↗

$0.02

Inference provider
stepFun
Platform
StepFun
Unit
cache_read_tokens_per_million
Currency
USD
Billing conditions
Billing label
Cached input / 1M tokens

step-3.5-flash ↗

$0.10

Inference provider
stepFun
Platform
StepFun
Unit
input_tokens_per_million
Currency
USD
Billing conditions
Billing label
Input / 1M tokens (cache miss)

step-3.5-flash ↗

$0.30

Inference provider
stepFun
Platform
StepFun
Unit
output_tokens_per_million
Currency
USD
Billing conditions
Billing label
Output / 1M tokens

Provider specifications & access

1–2 / 2

Step 3.5 Flash ↗

source model id
step-3.5-flash
hosting
stepfun-global-hosted
platforms
  • StepFun
context length
256000
architecture
input_modalities
  • text
output_modalities
  • text
capabilities
  • promptCaching
  • reasoning
  • streaming
  • structuredOutputs
  • toolCalling

Step-3.5-Flash ↗

source model id
stepfun-ai/Step-3.5-Flash
hosting
siliconflow-hosted
platforms
  • SiliconFlow
max output tokens
65536
architecture
input_modalities
  • text
output_modalities
  • text
capabilities
  • toolCalling

Source specification observations

1–3 / 3

siliconFlow ↗

field
context_length
value
262K
scope
source-model
source id
siliconFlow
confidence
1

stepFun ↗

field
context_length
value
256000
normalized value
256000
scope
source-model
source id
stepFun
confidence
1

siliconFlow ↗

field
max_output_tokens
value
66K
scope
source-model
source id
siliconFlow
confidence
1

Original sources

1–14 / 14

SiliconFlow · Model page ↗

StepFun · Model reference ↗

Hugging Face · Model repository ↗

DeepInfra · Model page ↗

SiliconFlow · Playground ↗

SiliconFlow · API documentation ↗

StepFun · Model documentation ↗

DeepInfra · Catalog ↗

Epoch AI · Catalog ↗

Hugging Face · Catalog ↗

Hugging Face · Catalog ↗

LMArena · Catalog ↗

SiliconFlow · Catalog ↗

StepFun · Catalog ↗

Source links

1–9 / 9

Explore more models

Models

Loading
EVIDENCE / COMPARISON

Benchmarks

Explore every benchmark

Loading

These charts contain model-linked results. The charts above also include unlinked source profiles. Compare scores within each evaluation. Model quality, usage, transparency and system performance measure different things.

Install TensorTrace