{ "meta": { "attribution": "Anchor Terminal (https://www.anchorterminal.com)", "attributionRequired": true, "credit": "CC BY 4.0, credit Anchor Terminal and link the page or API address you took it from (https://www.anchorterminal.com/terms/#data)", "docs": "https://www.anchorterminal.com/docs/", "generatedAt": "2026-11-11", "license": "CC-BY-4.0", "method": "https://www.anchorterminal.com/benchmark/", "methodology": "0.7", "openapi": "https://www.anchorterminal.com/openapi.json", "preview": false, "run": "2026-10-01", "runLabel": "October 2026 research run", "terms": "https://www.anchorterminal.com/terms/#data" }, "tool": { "slug": "step-3-7-flash", "name": "Step 3.7 Flash", "vendor": "StepFun", "vendorUrl": "https://platform.stepfun.ai", "kind": "weights", "category": "open-weights", "summary": "Step 3.7 Flash is StepFun's open-weight vision-language model, 198B parameters with about 11B active, with tool calling and a 256K context. Apache 2.0. Its 4-bit GGUF is 105 GB, just over what a 128 GB machine holds.", "url": "https://www.anchorterminal.com/tools/step-3-7-flash", "markdownUrl": "https://www.anchorterminal.com/tools/step-3-7-flash.md", "slimMarkdownUrl": "https://www.anchorterminal.com/tools/step-3-7-flash.min.md", "jsonUrl": "https://www.anchorterminal.com/api/v1/tools/step-3-7-flash.json", "dns": "step-3-7-flash.grade.anchorterminal.com", "repo": "https://huggingface.co/stepfun-ai/Step-3.7-Flash", "license": "Apache-2.0 (metadata and card; no LICENSE file)", "transports": [], "packages": [ { "registry": "huggingface", "name": "stepfun-ai/Step-3.7-Flash" }, { "registry": "huggingface", "name": "stepfun-ai/Step-3.7-Flash-GGUF" }, { "registry": "huggingface", "name": "stepfun-ai/Step-3.7-Flash-FP8" }, { "registry": "huggingface", "name": "stepfun-ai/Step-3.7-Flash-NVFP4" } ], "auth": "none", "authNotes": "The weights download from Hugging Face with no account, token or terms to accept. None of the four repositories is gated (checked 2026-11-10).", "pricing": "free", "pricingNotes": "Free to download and use under Apache 2.0. Running it yourself takes 109 GB at IQ4_XS with its projector, 129 GB in NVFP4 or 403 GB in BF16. StepFun's API serves it at $0.20 per million input tokens and $1.15 per million output, per the card, and DeepInfra and Featherless serve it too (checked 2026-11-10).", "priceSummary": "Free", "where": "local", "x402": { "level": "no", "evidence": "Weights you download and run yourself, so there is nothing to pay per call and no payment protocol (checked 2026-11-10).", "endpoints": [] }, "toolCount": null, "popularity": { "githubStars": null, "npmWeekly": null, "pypiWeekly": null, "asOf": "2026-11-10" }, "docsUrl": "https://huggingface.co/stepfun-ai/Step-3.7-Flash", "capabilities": [ "model.open-weights" ], "tags": [ "open-weights", "model", "moe", "multimodal", "vision", "reasoning", "tool-calling", "agentic", "long-context", "apache-2.0", "safetensors", "gguf", "fp8", "nvfp4", "custom-code", "self-hosted" ], "lastRelease": "2026-06-03", "graded": true, "anchor": { "graded": true, "score": 59.8, "grade": "C", "agentReady": false, "rank": 0, "ranked": false, "notRankedWhy": "An open-weight model, graded by the open-weight checklist and ranked among open-weight models, not against tools", "rankOf": 1518, "categoryRank": 73, "methodology": "0.7", "run": "2026-10-01", "scores": { "interface": 85, "reach": 70, "risk": 45, "upkeep": 39 }, "pending": [ "performance", "tasks" ], "breakdown": [ { "key": "capability", "name": "Capability", "weight": 35, "effectiveWeight": 0, "pending": true, "points": 0, "reason": "Pending for this listing. Nobody independent has published results for it in at least two areas yet, so its weight is shared across the other four and the total is capped at 61 until they do or our own runs start." }, { "key": "reach", "name": "Reach", "weight": 25, "effectiveWeight": 38.46, "score": 70, "points": 26.92, "reason": "Memory 5, the smallest build at 4 bits is StepFun's IQ4_XS GGUF, 104.99 GB with the 3.97 GB projector, and 108.96 GB × 1.2 = 130.8 GB, over what the reference machine can give a model and within 256 GB. The IQ3 and Q3 builds are below 4 bits and aren't read. Runtimes 20, Released vLLM 0.33.0, SGLang 0.5.23 and Transformers 5.22.2 have the step3p7 model, and llama.cpp b12242 has the architecture, with StepFun's own GGUF. Four at 5 each, and the card documents all four. Builds 15, StepFun publishes GGUF at five quantised precisions, an FP8 build and an NVFP4 build. Terms 15, Apache 2.0 in the card's metadata and its licence section, an OSI licence with nothing added. Download 10, the Hub API reports gated false for all four repositories on 10 November 2026, so the files download with no account and nothing to accept. Hosted 5, StepFun's own API serves it, per the card and our StepFun API listing, and Featherless and DeepInfra are live for it on Hugging Face. 5 + 20 + 15 + 15 + 10 + 5 = 70.", "checked": "2026-11-10" }, { "key": "risk", "name": "Risk", "weight": 20, "effectiveWeight": 30.77, "score": 45, "points": 13.85, "reason": "File safety 15, safetensors and GGUF only, no .bin, .pt or .pth files (25). Provenance 15, the files sit in the stepfun-ai organisation, which the card's StepFun platform and model page belong to (10). Safety evaluation 0, no safety results on the card. Agent guidance 0, the card has no limitations section and gives no guidance on prompt injection, instruction priority, refusals or safeguards. Licence risk 15, Apache 2.0 has no use policy, threshold or change clause. Security programme 0, no way to report a flaw on the card, and our StepFun API listing found no security.txt, bug bounty or disclosure policy. 15 + 15 + 0 + 0 + 15 + 0 = 45.", "checked": "2026-11-10" }, { "key": "interface", "name": "Agent interface", "weight": 10, "effectiveWeight": 15.38, "score": 85, "points": 13.08, "reason": "Chat template 20, chat_template.jinja in the repository (5,723 characters). Tool-calling format 25, the card names the parser to use in two runtimes, step3p5 in vLLM (with auto tool choice) and in SGLang, and both ship Step parsers in a release. Trained tool calling 20, the card presents tool orchestration as a main use, with multi-turn results, and the template renders every call in a turn. Structured output 0, Nothing on the card about JSON or schema-constrained output for the weights. Reasoning control 10, the template has a reasoning_effort setting and keeps reasoning_content, and the card names the step3p5 reasoning parser for vLLM and SGLang, which returns the reasoning apart from the answer. Context 10, 256K tokens per the card, and 262,144 positions in config.json. 20 + 25 + 20 + 0 + 10 + 10 = 85.", "checked": "2026-11-10" }, { "key": "upkeep", "name": "Documentation and upkeep", "weight": 10, "effectiveWeight": 15.38, "score": 39, "points": 6, "reason": "Model card 14, Intended use, the architecture, the parameters, the context and the language (in the metadata) are on the card. Training data 0, Nothing on the card about the training data. Last change 10, the last commit, on 3 June 2026, fixed the modelling code, 160 days before this check and within 180. Responsiveness 0, none of the 11 most recent discussions that aren't pull requests (29 May to 24 June 2026) has a reply from a stepfun-ai member, counted on 10 November by author flags and dates with a script, text not read. Version policy 5, Step-3.5-Flash stays in its own repositories and 3.7 came out in new ones, with no dated note on the older cards that we read. File health 10, Config, template and card agree on the architecture and the context, and June commits fixed the modelling code. 14 + 0 + 10 + 0 + 5 + 10 = 39.", "checked": "2026-11-10" } ], "assessment": { "date": "2026-11-10", "basis": "public evidence", "confidence": "low", "notes": { "capability": "Pending for this listing. Independent results were found in 0 areas (none), and Capability needs 2. Its weight is shared across the other four categories and the total is capped at 61 until there are more, or our own runs start.", "interface": "Chat template 20, chat_template.jinja in the repository (5,723 characters). Tool-calling format 25, the card names the parser to use in two runtimes, step3p5 in vLLM (with auto tool choice) and in SGLang, and both ship Step parsers in a release. Trained tool calling 20, the card presents tool orchestration as a main use, with multi-turn results, and the template renders every call in a turn. Structured output 0, Nothing on the card about JSON or schema-constrained output for the weights. Reasoning control 10, the template has a reasoning_effort setting and keeps reasoning_content, and the card names the step3p5 reasoning parser for vLLM and SGLang, which returns the reasoning apart from the answer. Context 10, 256K tokens per the card, and 262,144 positions in config.json. 20 + 25 + 20 + 0 + 10 + 10 = 85.", "reach": "Memory 5, the smallest build at 4 bits is StepFun's IQ4_XS GGUF, 104.99 GB with the 3.97 GB projector, and 108.96 GB × 1.2 = 130.8 GB, over what the reference machine can give a model and within 256 GB. The IQ3 and Q3 builds are below 4 bits and aren't read. Runtimes 20, Released vLLM 0.33.0, SGLang 0.5.23 and Transformers 5.22.2 have the step3p7 model, and llama.cpp b12242 has the architecture, with StepFun's own GGUF. Four at 5 each, and the card documents all four. Builds 15, StepFun publishes GGUF at five quantised precisions, an FP8 build and an NVFP4 build. Terms 15, Apache 2.0 in the card's metadata and its licence section, an OSI licence with nothing added. Download 10, the Hub API reports gated false for all four repositories on 10 November 2026, so the files download with no account and nothing to accept. Hosted 5, StepFun's own API serves it, per the card and our StepFun API listing, and Featherless and DeepInfra are live for it on Hugging Face. 5 + 20 + 15 + 15 + 10 + 5 = 70.", "risk": "File safety 15, safetensors and GGUF only, no .bin, .pt or .pth files (25). Provenance 15, the files sit in the stepfun-ai organisation, which the card's StepFun platform and model page belong to (10). Safety evaluation 0, no safety results on the card. Agent guidance 0, the card has no limitations section and gives no guidance on prompt injection, instruction priority, refusals or safeguards. Licence risk 15, Apache 2.0 has no use policy, threshold or change clause. Security programme 0, no way to report a flaw on the card, and our StepFun API listing found no security.txt, bug bounty or disclosure policy. 15 + 15 + 0 + 0 + 15 + 0 = 45.", "upkeep": "Model card 14, Intended use, the architecture, the parameters, the context and the language (in the metadata) are on the card. Training data 0, Nothing on the card about the training data. Last change 10, the last commit, on 3 June 2026, fixed the modelling code, 160 days before this check and within 180. Responsiveness 0, none of the 11 most recent discussions that aren't pull requests (29 May to 24 June 2026) has a reply from a stepfun-ai member, counted on 10 November by author flags and dates with a script, text not read. Version policy 5, Step-3.5-Flash stays in its own repositories and 3.7 came out in new ones, with no dated note on the older cards that we read. File health 10, Config, template and card agree on the architecture and the context, and June commits fixed the modelling code. 14 + 0 + 10 + 0 + 5 + 10 = 39." }, "sources": [ { "what": "model card", "url": "https://huggingface.co/stepfun-ai/Step-3.7-Flash", "seen": "2026-11-10" }, { "what": "gating, dates, parameter count, downloads, likes, file list and hashes", "url": "https://huggingface.co/api/models/stepfun-ai/Step-3.7-Flash?blobs=true", "seen": "2026-11-10" }, { "what": "StepFun's GGUF repository: files and sizes", "url": "https://huggingface.co/api/models/stepfun-ai/Step-3.7-Flash-GGUF?blobs=true", "seen": "2026-11-10" }, { "what": "the FP8 repository", "url": "https://huggingface.co/api/models/stepfun-ai/Step-3.7-Flash-FP8?blobs=true", "seen": "2026-11-10" }, { "what": "the NVFP4 repository", "url": "https://huggingface.co/api/models/stepfun-ai/Step-3.7-Flash-NVFP4?blobs=true", "seen": "2026-11-10" }, { "what": "inference providers, Featherless and DeepInfra live", "url": "https://huggingface.co/api/models/stepfun-ai/Step-3.7-Flash?expand[]=inferenceProviderMapping", "seen": "2026-11-10" }, { "what": "commit history", "url": "https://huggingface.co/api/models/stepfun-ai/Step-3.7-Flash/commits/main", "seen": "2026-11-10" }, { "what": "architecture, auto_map and positions", "url": "https://huggingface.co/stepfun-ai/Step-3.7-Flash/raw/main/config.json", "seen": "2026-11-10" }, { "what": "chat template", "url": "https://huggingface.co/stepfun-ai/Step-3.7-Flash/raw/main/chat_template.jinja", "seen": "2026-11-10" }, { "what": "discussions, replies counted by script from authors, dates and member flags", "url": "https://huggingface.co/api/models/stepfun-ai/Step-3.7-Flash/discussions", "seen": "2026-11-10" }, { "what": "the organisation's other repositories", "url": "https://huggingface.co/api/models?author=stepfun-ai", "seen": "2026-11-10" }, { "what": "vLLM v0.33.0, step3p7.py and the Step parsers", "url": "https://raw.githubusercontent.com/vllm-project/vllm/v0.33.0/vllm/model_executor/models/step3p7.py", "seen": "2026-11-10" }, { "what": "SGLang v0.5.23, srt/models/step3p7.py and the parser names", "url": "https://raw.githubusercontent.com/sgl-project/sglang/v0.5.23/python/sglang/srt/models/step3p7.py", "seen": "2026-11-10" }, { "what": "Transformers v5.22.2 step3p7", "url": "https://github.com/huggingface/transformers/tree/v5.22.2/src/transformers/models/step3p7", "seen": "2026-11-10" }, { "what": "llama.cpp b12242, the Step architecture (shallow clone)", "url": "https://raw.githubusercontent.com/ggml-org/llama.cpp/b12242/src/llama-arch.cpp", "seen": "2026-11-10" }, { "what": "our StepFun API listing, for company facts", "url": "https://www.anchorterminal.com/tools/stepfun-api", "seen": "2026-11-10" } ], "openQuestions": [ "The card says it runs on a Ryzen AI Max+ 395 with 128 GB. Our calculator gives such a machine 96 GB for a model, and the smallest 4-bit build needs about 131 GB with its projector, or 126 GB without. The IQ3_XXS build (76 GB) would fit and is below our 4-bit line", "It's linked to the StepFun API listing with sameModel. That listing covers all of StepFun's models, not this one alone", "Reasoning control is 10 on the template's setting and the named parser. The card shows no request that sets it", "Capability is pending because no independent results were looked up on the day", "unchecked on 10 November: the model page on stepfun.com, the repository's Python files, the Docker images the card names, NVIDIA NIM, OpenRouter, Ollama's library and LM Studio's catalogue" ] }, "negative": 0, "verdict": "A capable open agent model with a clean licence, every kind of build and four released runtimes. Two things hold it back: it's a little too big for a 128 GB machine at 4 bits, whatever the card says, and StepFun has gone quiet on the repository since June. No safety results, limitations or data description.", "bestFor": "Teams with a GPU server or a provider account that want an open, long-context agent model with vision", "strengths": [ "Apache 2.0, ungated", "GGUF, FP8 and NVFP4 builds from StepFun", "Four released runtimes carry it, with Step parsers in vLLM and SGLang", "A 256K context with images in", "Hosted by StepFun and two providers" ], "weaknesses": [ "About 131 GB to run at 4 bits with images, over a 128 GB machine's share", "Every serve command on the card runs the repository's code", "No reply from StepFun on the repository's discussions", "Not changed since 3 June 2026", "No limitations, safety results or training data on the card" ], "agentNotes": [ "On a 128 GB machine the 4-bit GGUF won't fit with a working context. Use StepFun's API or a provider, or a server", "On vLLM or SGLang pass `--tool-call-parser step3p5 --reasoning-parser step3p5`", "The card's commands pass trust-remote-code. Released Transformers 5.22.2 has the model built in if you'd rather not run the repository's code", "For llama.cpp load the mmproj file from StepFun's GGUF repository for image input", "StepFun's API has two hosts. Use the one where your key was issued, `api.stepfun.ai` or `api.stepfun.com`", "The same model is on our StepFun API listing, with prices and terms" ], "metrics": { "kind": "local", "measured": false }, "reviewCount": 0, "avgRating": 0, "history": [ { "basis": "public evidence", "confidence": "low", "grade": "C", "methodology": "0.7", "pending": [ "performance", "tasks" ], "run": "2026-10-01", "runLabel": "October 2026 research run", "score": 59.8 } ], "editorialScores": { "interface": 85, "reach": 70, "risk": 45, "upkeep": 39 }, "provenanceScore": 58, "checked": { "categories": { "interface": "2026-11-10", "reach": "2026-11-10", "risk": "2026-11-10", "upkeep": "2026-11-10" }, "last": "2026-11-10", "oldest": "2026-11-10", "stale": false, "staleAfterDays": 90 }, "scoreHistory": [ { "date": "2026-11-10", "grade": "C", "kind": "listed", "method": "0.7", "score": 59.8 } ] }, "connect": { "install": "pip install \"sglang\u003e=0.5.23\"\nsglang serve --model-path stepfun-ai/Step-3.7-Flash-FP8 --tp 8 --ep 4 --reasoning-parser step3p5 --tool-call-parser step3p5 --enable-multimodal --trust-remote-code", "http": "# SGLang and vLLM serve an OpenAI-compatible API. StepFun's own API is at https://api.stepfun.ai/v1\n# (https://api.stepfun.com/v1 for keys issued in China)" }, "letme": { "capability": "https://letme.dev/model.open-weights", "tool": "https://letme.dev/step-3-7-flash" }, "sameModel": [ "stepfun-api" ], "weights": { "trainedBy": "StepFun", "params": "198B", "active": "11B", "context": 262144, "goodFor": "Agent work that mixes images, search and code, on a GPU server or through StepFun's API. It's a few GB too big for a 128 GB machine at 4 bits.", "runtimes": [ "llama.cpp", "vLLM", "SGLang", "Transformers" ], "builds": [ { "name": "Step-3.7-Flash-GGUF IQ4_XS", "url": "https://huggingface.co/stepfun-ai/Step-3.7-Flash-GGUF", "who": "publisher", "precision": "iq4_xs", "sizeGB": 108.96, "memoryGB": 192, "format": "gguf", "runsOn": [ "llama.cpp" ], "note": "104.99 GB with the 3.97 GB F16 projector" }, { "name": "Step-3.7-Flash-GGUF Q4_K_S", "url": "https://huggingface.co/stepfun-ai/Step-3.7-Flash-GGUF", "who": "publisher", "precision": "q4_k_s", "sizeGB": 115.47, "memoryGB": 192, "format": "gguf", "runsOn": [ "llama.cpp" ], "note": "111.50 GB with the 3.97 GB F16 projector" }, { "name": "Step-3.7-Flash-NVFP4", "url": "https://huggingface.co/stepfun-ai/Step-3.7-Flash-NVFP4", "who": "publisher", "precision": "nvfp4", "sizeGB": 129.25, "memoryGB": 192, "format": "safetensors", "runsOn": [ "vLLM", "SGLang" ], "note": "StepFun's 4-bit build for NVIDIA GPUs" }, { "name": "Step-3.7-Flash-FP8", "url": "https://huggingface.co/stepfun-ai/Step-3.7-Flash-FP8", "who": "publisher", "precision": "fp8", "sizeGB": 212.53, "memoryGB": 256, "format": "safetensors", "runsOn": [ "vLLM", "SGLang" ], "note": "StepFun's FP8 build" }, { "name": "Step-3.7-Flash-GGUF Q8_0", "url": "https://huggingface.co/stepfun-ai/Step-3.7-Flash-GGUF", "who": "publisher", "precision": "q8_0", "sizeGB": 213.39, "memoryGB": 512, "format": "gguf", "runsOn": [ "llama.cpp" ], "note": "209.42 GB with the 3.97 GB F16 projector" }, { "name": "Step-3.7-Flash (BF16 safetensors)", "url": "https://huggingface.co/stepfun-ai/Step-3.7-Flash", "who": "publisher", "precision": "bf16", "sizeGB": 402.73, "memoryGB": 512, "format": "safetensors", "runsOn": [ "vLLM", "SGLang", "Transformers" ], "note": "26 files" } ], "run": "llama-server -hf stepfun-ai/Step-3.7-Flash-GGUF:IQ4_XS --jinja", "runBuild": "Step-3.7-Flash-GGUF IQ4_XS", "checked": "2026-11-10", "minMemoryGB": 192 }, "family": { "lab": "StepFun", "family": "Step 3", "url": "https://www.anchorterminal.com/open-weights/#stepfun", "others": [ { "name": "Step-3.5-Flash", "url": "https://huggingface.co/stepfun-ai/Step-3.5-Flash" }, { "name": "Step3-VL-10B", "url": "https://huggingface.co/stepfun-ai/Step3-VL-10B" } ] }, "notable": [ "10,411 downloads in the last 30 days and 463 likes on Hugging Face on 10 November 2026 (https://huggingface.co/api/models/stepfun-ai/Step-3.7-Flash)", "198B parameters with about 11B active per token and a 1.8B vision encoder, per the card. The files hold 201.4B (https://huggingface.co/stepfun-ai/Step-3.7-Flash)", "The card says it runs on machines with at least 128 GB of unified memory, naming the Ryzen AI Max+ 395. By our calculator its smallest 4-bit GGUF needs about 131 GB with the projector, more than such a machine can give a model (https://huggingface.co/stepfun-ai/Step-3.7-Flash-GGUF)", "StepFun publishes GGUF at five quantised precisions, FP8 and NVFP4 builds, each in its own repository (https://huggingface.co/stepfun-ai/Step-3.7-Flash)", "The card reports 67.1 on ClawEval-1.1 and 56.3 on SWE-Bench PRO, and says Terminal-Bench 2.1 (59.5) and GDPVal-AA (45.8) show room to improve. These are StepFun's numbers (https://huggingface.co/stepfun-ai/Step-3.7-Flash)", "None of the 11 most recent discussions has a reply from a StepFun member, and the repository hasn't changed since 3 June 2026 (https://huggingface.co/api/models/stepfun-ai/Step-3.7-Flash/discussions)" ], "area": "models", "details": [ { "label": "Surface graded", "value": "The open weights at huggingface.co/stepfun-ai/Step-3.7-Flash and the GGUF, FP8 and NVFP4 repositories beside it, read from the model card, the files and their metadata, the config, the chat template, the commit history and the runtimes' own code at their latest releases on 10 November 2026. StepFun's company facts come from our StepFun API listing. The model page on stepfun.com, the repository's Python files and the pre-built Docker images weren't read. Nothing was downloaded or run" }, { "label": "Model", "value": "201,365,316,160 parameters in the files. The card says 198B: a 196B sparse mixture-of-experts language model with about 11B active per token, 45 layers, and a 1.8B vision encoder. Step3p7ForConditionalGeneration" }, { "label": "Files", "value": "26 safetensors files in BF16 (402.7 GB), chat_template.jinja, config and tokenizer files and four Python modules. No pickle files. The GGUF repository has IQ3_XXS to BF16 with a projector and MTP files. Hugging Face lists a SHA-256 for the large files" }, { "label": "Licence", "value": "Apache 2.0 in the metadata of every repository and in the card's licence section. There is no LICENSE file" }, { "label": "Runtimes", "value": "The card documents vLLM and SGLang, each from a pre-built image, Transformers 5.0 or later with trust_remote_code for checking, and llama.cpp. Released vLLM 0.33.0, SGLang 0.5.23 and Transformers 5.22.2 have the step3p7 model, and llama.cpp b12242 has the architecture, with StepFun's own GGUF" }, { "label": "Context", "value": "256K tokens per the card, and 262,144 positions in config.json. The tokenizer config says 131,072, and the card's NVFP4 example serves at 8,192" }, { "label": "Tools and reasoning", "value": "The card's vLLM and SGLang commands name the step3p5 tool-call parser, with auto tool choice on vLLM, and the step3p5 reasoning parser. The chat template takes tools and a reasoning_effort setting and keeps reasoning_content. The card describes tool orchestration results and doesn't show the setting in a request" }, { "label": "Hardware", "value": "The card says data-centre GPUs for production, and for local use machines with at least 128 GB of unified memory: NVIDIA DGX Station, AMD Ryzen AI Max+ 395 systems and Mac Studio or MacBook Pro. Its serve commands use 8 GPUs for BF16 and FP8 and 4 for NVFP4" }, { "label": "Training data", "value": "Not described on the card" } ], "taskCosts": { "hosted": [ { "openrouterId": "stepfun/step-3.7-flash", "name": "Step 3.7 Flash", "source": "OpenRouter, $0.20 in and $1.15 out per 1M tokens", "usd": { "bugfix": 0.03, "research": 0.52, "website": 0.16 } } ], "note": "Each figure is the same tokens priced at this model's rates, research report (134 fresh input, 207k cache writes, 10.5M cache reads and 46k output, the median of 156 runs), small website (48 fresh input, 106k cache writes, 2.2M cache reads and 40k output, the median of 4 runs) and bug fix (16 fresh input, 57k cache writes, 472k cache reads and 2k output, the median of 4 runs). We measured the tokens with Claude Opus 5.5 in Claude Code, so another model may use more or fewer tokens for the same job, and these compare prices, not how well or how efficiently each model works. Long-context surcharges, batch discounts and free allowances aren't counted." }, "provenance": { "legalEntity": "STEPFUN TECHNOLOGY (HK) LIMITED", "domain": "stepfun.ai", "domainRegistered": "", "endpointOnVendorDomain": null, "terms": "", "privacy": "", "statusPage": "", "changelog": "https://huggingface.co/stepfun-ai/Step-3.7-Flash/commits/main", "securityTxt": "none", "checked": "2026-11-10", "notes": [ "StepFun's legal entity, domain and security.txt are carried from our StepFun API listing.", "Nothing is hosted under this listing, so Apache 2.0 stands in for terms of service and no privacy policy is scored. The StepFun API that serves this model has its own listing.", "The changelog is the commit history, 23 commits from 23 May to 3 June 2026." ], "score": 58, "checks": [ { "check": "Legal entity named", "value": "STEPFUN TECHNOLOGY (HK) LIMITED", "points": 20, "max": 20, "state": "ok" }, { "check": "Domain age", "value": "stepfun.ai, no registry record we could read", "points": 0, "max": 15, "state": "no" }, { "check": "Endpoint on the vendor's domain", "value": "no hosted endpoint", "points": 0, "max": 0, "state": "na" }, { "check": "Terms of service", "value": "nothing hosted, so the Apache-2.0 (metadata and card; no LICENSE file) licence stands in", "points": 10, "max": 10, "state": "ok" }, { "check": "Privacy policy", "value": "nothing hosted, not scored", "points": 0, "max": 0, "state": "na" }, { "check": "Status page", "value": "not found", "points": 0, "max": 10, "state": "no" }, { "check": "Changelog", "value": "published", "points": 10, "max": 10, "state": "ok" }, { "check": "security.txt", "value": "not found", "points": 0, "max": 10, "state": "no" }, { "check": "People named", "value": "2 named with their role, none with a public profile", "points": 7, "max": 10, "state": "part" }, { "check": "Support route", "value": "help pages", "points": 5, "max": 5, "state": "ok" } ] }, "humans": { "slug": "step-3-7-flash", "checked": "2026-11-10", "company": "StepFun", "people": [ { "name": "Daxin Jiang", "role": "Co-founder and CEO", "founder": true, "bio": "Daxin Jiang is co-founder and CEO of StepFun, a Shanghai AI company founded in April 2023 that builds the Step series of language, multimodal and speech models. He was previously a global vice president at Microsoft, and StepFun's founding team came from Microsoft.", "bioSources": [ "https://en.wikipedia.org/wiki/StepFun", "https://zh.wikipedia.org/wiki/%E9%98%B6%E8%B7%83%E6%98%9F%E8%BE%B0" ], "page": "/people/daxin-jiang/" }, { "name": "Yibo Zhu", "role": "Co-founder; head of systems", "founder": true, "bio": "Yibo Zhu is a co-founder of StepFun, the Shanghai AI company behind the Step model series, where he leads systems work. He has experience building and running large-scale systems and previously worked at ByteDance and Google.", "bioSources": [ "https://en.wikipedia.org/wiki/StepFun", "https://zh.wikipedia.org/wiki/%E9%98%B6%E8%B7%83%E6%98%9F%E8%BE%B0" ], "page": "/people/yibo-zhu/" } ], "peopleNote": "StepFun's own site (stepfun.com, including /about) only renders in a browser and was left unread; the API platform at platform.stepfun.ai names no people and its Step 5 Preview blog post carries no byline. English Wikipedia (read 28 October 2026) lists the founders as Jiang Daxin and Zhu Yibo; Chinese Wikipedia says Jiang Daxin founded the company and is CEO, and names Zhu Yibo as head of systems and Jiao Binxing as head of data. KrASIA calls Jiang Daxin co-founder and CEO. Founder status and roles rest on Wikipedia and the press. yibozhu.com did not answer for robots.txt and was not read.", "companySocial": { "linkedin": "https://www.linkedin.com/company/stepfun-ai", "x": "https://x.com/StepFun_ai" }, "support": { "urls": [ "https://platform.stepfun.ai/developer-community" ] }, "sources": [ "https://cn.linkedin.com/in/daxin-jiang-00006124", "https://yibozhu.com/", "https://www.linkedin.com/in/yibo-zhu-66494750", "https://www.linkedin.com/company/stepfun-ai", "https://x.com/StepFun_ai", "https://platform.stepfun.ai/developer-community", "https://platform.stepfun.com" ], "ownership": { "status": "private", "checked": "2026-10-29", "sources": [ "https://technode.com/2026/01/26/shanghai-ai-unicorn-stepfun-raises-over-718-million-in-b-round/", "https://kr-asia.com/as-ai-consolidates-what-makes-stepfun-worth-a-rmb-5-billion-raise", "https://www.thestandard.com.hk/finance/article/331515/Stepfun-Chinas-AI-Six-Tigers-finishes-new-US25b-funding-round-for-HK-IPO" ] }, "investors": [ { "name": "HongShan", "kind": "vc", "rounds": [ "First round" ], "source": "https://kr-asia.com/as-ai-consolidates-what-makes-stepfun-worth-a-rmb-5-billion-raise", "quote": "StepFun’s first round therefore relied heavily on the personal networks of its co-founders, including HongShan, Qiming Venture Partners, and IDG Capital.", "page": "/investors/hongshan/" }, { "name": "Qiming Venture Partners", "kind": "vc", "rounds": [ "First round", "Series B+, January 2026" ], "source": "https://technode.com/2026/01/26/shanghai-ai-unicorn-stepfun-raises-over-718-million-in-b-round/", "quote": "Existing shareholders Tencent Holdings, Qiming Venture Partners, and 5Y Capital also participated in follow-on investments.", "page": "/investors/qiming-venture-partners/" }, { "name": "IDG Capital", "kind": "vc", "rounds": [ "First round" ], "source": "https://kr-asia.com/as-ai-consolidates-what-makes-stepfun-worth-a-rmb-5-billion-raise", "quote": "StepFun’s first round therefore relied heavily on the personal networks of its co-founders, including HongShan, Qiming Venture Partners, and IDG Capital.", "page": "/investors/idg-capital/" }, { "name": "5Y Capital", "kind": "vc", "rounds": [ "Second round", "Series B+, January 2026" ], "source": "https://technode.com/2026/01/26/shanghai-ai-unicorn-stepfun-raises-over-718-million-in-b-round/", "quote": "Existing shareholders Tencent Holdings, Qiming Venture Partners, and 5Y Capital also participated in follow-on investments.", "page": "/investors/5y-capital/" }, { "name": "Shunwei Capital", "kind": "vc", "rounds": [ "Second round" ], "source": "https://kr-asia.com/as-ai-consolidates-what-makes-stepfun-worth-a-rmb-5-billion-raise", "quote": "its second round, which drew a broader investor base, including USD-denominated funds managed by 5Y Capital and Shunwei Capital, as well as strategic investors such as Lenovo Capital.", "page": "/investors/shunwei-capital/" }, { "name": "Lenovo Capital", "kind": "corporate", "rounds": [ "Second round" ], "source": "https://kr-asia.com/as-ai-consolidates-what-makes-stepfun-worth-a-rmb-5-billion-raise", "quote": "its second round, which drew a broader investor base, including USD-denominated funds managed by 5Y Capital and Shunwei Capital, as well as strategic investors such as Lenovo Capital.", "page": "/investors/lenovo-capital/" }, { "name": "Shanghai State-owned Capital Investment", "kind": "other", "rounds": [ "Series B, launched mid-2024" ], "source": "https://kr-asia.com/as-ai-consolidates-what-makes-stepfun-worth-a-rmb-5-billion-raise", "quote": "Investors included Shanghai State-owned Capital Investment and affiliated funds, Hong Kong Investment Corporation, Tencent, and returning shareholders.", "page": "/investors/shanghai-state-owned-capital-investment/" }, { "name": "Hong Kong Investment Corporation", "kind": "other", "rounds": [ "Series B, launched mid-2024" ], "source": "https://kr-asia.com/as-ai-consolidates-what-makes-stepfun-worth-a-rmb-5-billion-raise", "quote": "Investors included Shanghai State-owned Capital Investment and affiliated funds, Hong Kong Investment Corporation, Tencent, and returning shareholders.", "page": "/investors/hong-kong-investment-corporation/" }, { "name": "Tencent", "kind": "corporate", "rounds": [ "Series B, launched mid-2024", "Series B+, January 2026" ], "source": "https://technode.com/2026/01/26/shanghai-ai-unicorn-stepfun-raises-over-718-million-in-b-round/", "quote": "Existing shareholders Tencent Holdings, Qiming Venture Partners, and 5Y Capital also participated in follow-on investments.", "page": "/investors/tencent/" }, { "name": "Shanghai SDIC Leading Fund", "kind": "other", "rounds": [ "Series B+, January 2026" ], "source": "https://technode.com/2026/01/26/shanghai-ai-unicorn-stepfun-raises-over-718-million-in-b-round/", "quote": "The round saw participation from Shanghai SDIC Leading Fund, China Life Private Equity Investment, Pudong Venture Capital, Xuhui Capital, Wuxi Liangxi Fund, Xiamen ITG Group, and Huaqin Technology, among other industrial investors.", "page": "/investors/shanghai-sdic-leading-fund/" }, { "name": "China Life Private Equity Investment", "kind": "other", "rounds": [ "Series B+, January 2026" ], "source": "https://technode.com/2026/01/26/shanghai-ai-unicorn-stepfun-raises-over-718-million-in-b-round/", "quote": "The round saw participation from Shanghai SDIC Leading Fund, China Life Private Equity Investment, Pudong Venture Capital, Xuhui Capital, Wuxi Liangxi Fund, Xiamen ITG Group, and Huaqin Technology, among other industrial investors.", "page": "/investors/china-life-private-equity-investment/" }, { "name": "Pudong Venture Capital", "kind": "other", "rounds": [ "Series B+, January 2026" ], "source": "https://technode.com/2026/01/26/shanghai-ai-unicorn-stepfun-raises-over-718-million-in-b-round/", "quote": "The round saw participation from Shanghai SDIC Leading Fund, China Life Private Equity Investment, Pudong Venture Capital, Xuhui Capital, Wuxi Liangxi Fund, Xiamen ITG Group, and Huaqin Technology, among other industrial investors.", "page": "/investors/pudong-venture-capital/" }, { "name": "Xuhui Capital", "kind": "other", "rounds": [ "Series B+, January 2026" ], "source": "https://technode.com/2026/01/26/shanghai-ai-unicorn-stepfun-raises-over-718-million-in-b-round/", "quote": "The round saw participation from Shanghai SDIC Leading Fund, China Life Private Equity Investment, Pudong Venture Capital, Xuhui Capital, Wuxi Liangxi Fund, Xiamen ITG Group, and Huaqin Technology, among other industrial investors.", "page": "/investors/xuhui-capital/" }, { "name": "Wuxi Liangxi Fund", "kind": "other", "rounds": [ "Series B+, January 2026" ], "source": "https://technode.com/2026/01/26/shanghai-ai-unicorn-stepfun-raises-over-718-million-in-b-round/", "quote": "The round saw participation from Shanghai SDIC Leading Fund, China Life Private Equity Investment, Pudong Venture Capital, Xuhui Capital, Wuxi Liangxi Fund, Xiamen ITG Group, and Huaqin Technology, among other industrial investors.", "page": "/investors/wuxi-liangxi-fund/" }, { "name": "Xiamen ITG Group", "kind": "corporate", "rounds": [ "Series B+, January 2026" ], "source": "https://technode.com/2026/01/26/shanghai-ai-unicorn-stepfun-raises-over-718-million-in-b-round/", "quote": "The round saw participation from Shanghai SDIC Leading Fund, China Life Private Equity Investment, Pudong Venture Capital, Xuhui Capital, Wuxi Liangxi Fund, Xiamen ITG Group, and Huaqin Technology, among other industrial investors.", "page": "/investors/xiamen-itg-group/" }, { "name": "Huaqin Technology", "kind": "corporate", "rounds": [ "Series B+, January 2026", "US$2.5 billion round, reported May 2026" ], "source": "https://technode.com/2026/01/26/shanghai-ai-unicorn-stepfun-raises-over-718-million-in-b-round/", "quote": "The round saw participation from Shanghai SDIC Leading Fund, China Life Private Equity Investment, Pudong Venture Capital, Xuhui Capital, Wuxi Liangxi Fund, Xiamen ITG Group, and Huaqin Technology, among other industrial investors.", "page": "/investors/huaqin-technology/" } ] }, "pageJsonUrl": "https://www.anchorterminal.com/tools/step-3-7-flash.json" } }