Qwen3-Coder-30B-A3B-Instruct vs Step 3.7 Flash

Qwen3-Coder-30B-A3B-Instruct 90

Memory 25 of 30, LM Studio's MLX 4-bit build is 17.2 GB and Ollama's 4-bit build 19 GB, 21 to 23 GB with a fifth added, which fits 24 GB. Runtimes 25, released Transformers, vLLM, SGLang and llama.cpp run the qwen3_moe architecture, vLLM and SGLang carry a qwen3_coder parser, Ollama's released build pins a llama.cpp that has qwen3moe, and LM Studio's catalogue carries it. Builds 10 of 15, Alibaba's only quantised build is FP8. Terms 15, Apache 2.0 with nothing added. Download 10, neither repository is gated. Hosted 5, Featherless AI on Hugging Face (live on 4 November 2026) plus SiliconFlow and Amazon Bedrock on OpenRouter (per 21 October, not re-read) serve it. 25 + 25 + 10 + 15 + 10 + 5 = 90.

Step 3.7 Flash 70

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.

Qwen3-Coder-30B-A3B-Instruct 55

File safety 25, safetensors weights, and the shipped parser script runs neither to load the model nor to build prompts. Provenance 15, the files are in the Qwen organisation that Qwen's GitHub links to, with a checksum on every file. Safety evaluation 0, none published. Agent guidance 0, nothing on prompt injection, refusals or safeguards, though it's sold for agents that run commands. Licence risk 15, plain Apache 2.0 that hasn't changed. Security programme 0, no security.txt, disclosure policy or security contact for Qwen was found, and Alibaba's group security centre couldn't be checked, so it's scored as absent. Files unchanged on 4 November, and QwenLM/Qwen3-Coder still has no SECURITY.md. 25 + 15 + 0 + 0 + 15 + 0 = 55.

Step 3.7 Flash 45

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.

Qwen3-Coder-30B-A3B-Instruct 75

Chat template 20, chat_template.jinja in the repository. Tool-calling format 25, the card shows how tools are declared, the README names the parser for SGLang and vLLM, and the repository ships it. Trained tool calling 20, the template renders several calls a turn and the training used agent reinforcement learning. Structured output 0, not documented. Reasoning control 0, the model has no thinking. Context 10, 262,144 tokens native, with the YaRN extension not counted. Template, parser and config unchanged on 4 November. 20 + 25 + 20 + 0 + 0 + 10 = 75.

Step 3.7 Flash 85

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.

Qwen3-Coder-30B-A3B-Instruct 31

Model card 11 of 20, intended use, architecture, parameters and context are there, but limitations, languages and a knowledge cut-off aren't. Training data 5, the blog describes the mix in general terms without a cut-off. Last change 5, config.json was corrected on 3 December 2025, 336 days before the 4 November 2026 check. Responsiveness 0, counted by script on 4 November, none of 26 discussions has a reply from Alibaba's Qwen staff, and the one Hub-flagged member who replied is the independent quantiser bartowski. Version policy 5, a first release. File health 5, the files now agree with the card, but reports about the parser and an attention error wait without a reply. 11 + 5 + 5 + 0 + 5 + 5 = 31.

Step 3.7 Flash 39

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.