Step 3.5 Flash
StepFunfamily · Step3.5huggingface.co/stepfun-ai/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....
Updated 1 h ago · first seen 11 Sept 2026
model_01M294WW5QV0E1S0GS8B5HA1CP
Overview
Identity
- Canonical model
- Yesidentity confidence: mediumOne row per real model release. Artifacts (checkpoints, quantisations, conversions) and folded evaluation variants point here.
- Official checkpoints
- official_checkpoints = hf_repo identifiers carried by the model itself; artifacts are separate entities pointing here through canonical_id.
- Artifacts
- 2 quantizations0 official · 2 third-partySeparate entities (checkpoint · quantization · conversion · packaging) pointing to this model through canonical_id.
- Provider deployments
- 1
- API aliases
- step-3-5-flashstep-3-5-flash-0202stepfun/step-3.5-flashIdentifiers under which providers and evaluators refer to this model.
- Folded evaluation variants
- 2Effort / thinking variants (…-high, …-non-reasoning) are result configurations of this model, not separate models. Their old URLs redirect here.
Openness
Open weights— weights downloadable under Apache-2.0; commercial use allowed; redistribution allowed; derivatives allowed; 4 dimensions unknown.
Weights downloadable under a permissive or Creative Commons licence allowing commercial use; code or data may be missing.
Weights
Yes
Inference code
—
Training code
—
Training data
—
Dataset
—
Commercial use
Yes
Redistribution
Yes
Derivatives
Yes
Licence: Apache License 2.0 (permissive · SPDX Apache-2.0)
dimensions marked null are unknown, not false
Key facts
- Release date
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 4 d agomedium
- Status
Source:Artificial AnalysisT2observed 5 d agomedium
- Model card
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 4 d agomedium
- Openrouter id
Source:OpenRouter public model & pricing listingT2observed 5 d agomedium
Architecture
- Architecture
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 4 d agomedium
- Model type
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 4 d agomedium
- Parameters
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 4 d agomedium
- Weights dtype
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 4 d agomedium
- File size
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 4 d agomedium
- Library name
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 4 d agomedium
- Pipeline tag
Source:Hugging Face Hub (public pages, model cards, papers)T2observed 4 d agomedium
- Hugging Face repo
Source:OpenRouter public model & pricing listingT2observed 1 h agomedium
Capabilities
Modalities
- Modalities
- text
- Input
- text
- Output
- text
Capabilities
Tool calling
Yes
OpenRouter public model & pricing listing · T2
Structured output
No
OpenRouter public model & pricing listing · T2
Reasoning
Yes
OpenRouter public model & pricing listing · T2
Vision
Unavailable
Audio
Unavailable
Fine-tuning available
Unavailable
- Context window
Source:OpenRouter public model & pricing listingT2observed 5 d agomedium
- Max output
Source:OpenRouter public model & pricing listingT2observed 5 d agomedium
Benchmarks24
Compare with another model →Comparable same task and conditions · Partially comparable same task, conditions differ (effort, temperature, judge) · Not comparable different variant or metric
Current rows only, grouped by benchmark → canonical metric → comparability group (task configuration). Effort variants folded into this model appear as rows of the same group. 24 current rows in total. “vs leader” compares with the current leader of the benchmark's primary group only; other groups are not directly comparable. Comparability rules →
Providers & Pricing1
All offers in the price terminal →USD per 1M tokens as published by each provider; native units (per-request fees, flex/priority tiers) are kept verbatim. Rows are append-only — every price change is kept in the history below. Cost of a workload →
Price history
Output price · USD / 1M tokens 1 provider
- OpenRouter
- OpenRouterfirst observed $0.3011 Sept 2026
Input price · USD / 1M tokens 1 provider
- OpenRouter
- OpenRouterfirst observed $0.1011 Sept 2026
Hardware fit37
Assumptions (7)
- Estimated, not measured: weights = parameters × bytes/param × 1.15 runtime overhead (or the observed artifact file size when one is recorded).
- bytes/param: 4bit = 0.5, 8bit = 1.0, fp16 = 2.0 (uniform quantization, no per-layer exceptions).
- KV cache: 2 × layers × kv_heads × head_dim × 2 bytes × context × batch when the architecture is known; otherwise 0.5 GB per 8 192 tokens (× batch), independent of architecture (GQA/MLA models need less).
- A model 'fits' when the estimate is at most the device memory minus 2 GB reserved for the OS and framework.
- Mixture-of-experts models are estimated on total parameters (all experts must be resident); active parameters are ignored.
- Device memory uses the largest configuration when several are listed (e.g. Apple silicon tiers).
- Multi-GPU: device memories are summed; interconnect bandwidth, tensor-parallel replication and pipeline bubbles are not modelled.
Lineage
Open in Graph →- descendant: Step-3.5-Flash-REAP-121B-A11B
- descendant: Step-3.5-Flash-REAP-149B-A11B
Versions & Artifacts2
Version history
Context window2 changes
11 Sept 2026→11 Sept 2026→12 Sept 2026current
Openness1 change
11 Sept 2026→11 Sept 2026current
Licensefirst observation only
12 Sept 2026current
Max outputfirst observation only
11 Sept 2026current
Parametersfirst observation only
12 Sept 2026current
Statusfirst observation only
11 Sept 2026current
Each hop is a claim: click a value for its source, tier and observation time. Nothing is overwritten — a new observation closes the previous claim.
Artifacts 2
quantizations 2
- stepfun-ai/Step-3.5-Flash-FP8StepFun · FP8 · BF16/F32/F8_E4M3208.6 GB
- stepfun-ai/Step-3.5-Flash-GGUF-Q4_K_SStepFun · GGUF—
Papers3
- arXiv:2602.10604Active35
- arXiv:2601.05593Active35
- arXiv:2507.19427Active35
Timeline19
Full timeline →Step 3.5 Flash: hf repo changed from stepfun-ai/Step-3.5-Flash to stepfun-ai/step-3.5-flash
Hugging Face repostepfun-ai/Step-3.5-Flash→stepfun-ai/step-3.5-flashhuggingfaceStep 3.5 Flash: hf repo changed from stepfun-ai/step-3.5-flash to stepfun-ai/Step-3.5-Flash
Hugging Face repostepfun-ai/step-3.5-flash→stepfun-ai/Step-3.5-FlashopenrouterStep 3.5 Flash: openness changed from proprietary to open-weights
Opennessproprietary→open-weightshuggingfaceStep 3.5 Flash: aa openness changed from proprietary to open-weights
Aa opennessproprietary→open-weightsartificial_analysisStep 3.5 Flash: release date changed from 2026-01-29 to 2026-02-01
Release date29 Jan 2026→1 Feb 2026huggingfaceStep 3.5 Flash scores 32.58% on Terminal-Bench
artificial_analysisStep 3.5 Flash scores 87.43% on τ²-bench
artificial_analysisStep 3.5 Flash scores 66.53% on IFBench
artificial_analysisStep 3.5 Flash scores 24.47% on Humanity's Last Exam
artificial_analysisStep 3.5 Flash scores 82.63% on GPQA Diamond
artificial_analysisStep 3.5 Flash scores 16.96 on Artificial Analysis Intelligence Index
artificial_analysisStep 3.5 Flash scores 27.27% on Terminal-Bench
artificial_analysisStep 3.5 Flash scores 94.44% on τ²-bench
artificial_analysisStep 3.5 Flash scores 64.56% on IFBench
artificial_analysisStep 3.5 Flash scores 21.13% on Humanity's Last Exam
artificial_analysisStep 3.5 Flash scores 83.13% on GPQA Diamond
artificial_analysisStep 3.5 Flash scores 16.62 on Artificial Analysis Intelligence Index
artificial_analysisStep 3.5 Flash: aa openness changed from open-weights to proprietary
Aa opennessopen-weights→proprietaryartificial_analysisStep 3.5 Flash: context length changed from 256000 to 262144
Context window256K tokens→262.1K tokensopenrouter
Change history167
Viewing AI Atlas as of 18 Jun 2026 — attributes exactly as the atlas knew them on that day; later corrections are not shown.
Back to today →Step 3.5 Flash was not yet in AI Atlas on 18 Jun 2026
Release daterelease_date2
Statusstatus2
Opennessopenness3
Licenselicense1
Architecturearchitecture1
Parametersparameter_count1
Context windowcontext_length4
Max outputmax_output_tokens1
Modalitiesmodalities1
Input modalitiesmodalities_input1
Output modalitiesmodalities_output1
Base modelbase_model1
File sizefile_size_gb1
Hugging Face repohf_repo41
Pipeline tagpipeline_tag1
Model cardmodel_card_url1
Downloadsmetric.downloads5
Likesmetric.likes6
Aa context windowaa_context_window1
Aa deprecatedaa_deprecated1
Aa opennessaa_openness32
Accessaccess1
Commercial use allowedcommercial_use_allowed1
Derivatives allowedderivatives_allowed1
Descriptiondescription1
Hf inference providershf_inference_providers1
Last modifiedlast_modified1
Library namelibrary_name1
License keylicense_key1
License rawlicense_raw1
Aa median output tokens per secondmetric.aa_median_output_tokens_per_second33
Downloads all timemetric.downloads_all_time5
Model typemodel_type1
Openrouter idopenrouter_id1
Openrouter listed atopenrouter_listed_at1
Reasoningreasoning2
Redistribution allowedredistribution_allowed1
Structured outputstructured_output1
Supported parameterssupported_parameters1
Tool callingtool_calling1
Weights availableweights_available1
Weights dtypeweights_dtype1
Claims are temporal and append-only: a new observation closes the previous claim (valid_to) instead of overwriting it. Conflicting claims from different sources are kept side by side and flagged — never averaged. Methodology →
Provenance
Attributed facts
44
Source tiers
T244
Freshest observation
1 h ago
Conflicts
None
Source documents 8
Tier 1 = official/primary, 2 = quality secondary, 3 = community, 4 = unverified. Every snapshot is archived; see all sources and the methodology.
Data quality (72/100) measures how well AI Atlas knows this entity — completeness, primary-source ratio, freshness, conflicts — never how good the model is. Methodology →