Nanbeige4.2-3B vs OTel-2.0-LLM-31B-IT

Nanbeige4.2-3B 85

Memory 30, the smallest build from Nanbeige is the FP8 repository (the 8-bit GPTQ one is the same size), 5.19 GB, and 5.19 GB × 1.2 = 6.2 GB, within 16 GB. bartowski's GGUF isn't read for this line. Runtimes 15, Released SGLang 0.5.23 has the model (srt/models/nanbeige.py) with its parsers, llama.cpp b12242 has the nanbeige architecture, and vLLM 0.33.0 names NanbeigeForCausalLM in its registry, served through its Transformers backend. Builds 15, Quantised builds from Nanbeige at two precisions, FP8 and 8-bit GPTQ, each in its own repository. Terms 15, Apache 2.0 in the card's metadata, an OSI licence, with no use policy or other terms on the card. Download 10, the Hub API reports gated false for the model and its FP8 and 8-bit repositories on 10 November 2026, so the files download with no account and nothing to accept. Hosted 0, Hugging Face maps no inference provider to it, and no Nanbeige or BOSS Zhipin API for it was found. 30 + 15 + 15 + 15 + 10 + 0 = 85.

OTel-2.0-LLM-31B-IT 50

Memory 10, the publisher's only build is the BF16 safetensors, 62.55 GB for the 15 shards the index names, and 62.55 GB × 1.2 = 75.1 GB, above 64 GB and inside the reference machine's GPU share. Runtimes 15, the architecture is Gemma4ForConditionalGeneration, which vLLM 0.33.0, SGLang 0.5.23 and Transformers 5.22.2 run from the safetensors files, three at 5 each. Builds 0, the publisher ships the BF16 weights only. Terms 15, Apache 2.0 in the card's metadata, an OSI licence, with nothing added on the card. Download 10, gated is false on 10 November 2026, so the files download with no account or terms. Hosted 0, Hugging Face maps one inference provider, Featherless AI (live), short of the two the line needs. 10 + 15 + 0 + 15 + 10 + 0 = 50.

Nanbeige4.2-3B 85

Chat template 20, A Jinja chat template in tokenizer_config.json (7,961 characters), with enable_thinking, preserve_thinking, tools and tool_call_format switches. Tool-calling format 25, the card says passing tools switches on the tool template and recommends tool_call_format xml, with json for compatibility, and its SGLang and vLLM commands name the nanbeige tool-call parser, which released SGLang 0.5.23 has. Trained tool calling 20, the card describes fine-tuning on tool-use environments and agent scaffolds and reinforcement learning on agent tasks, and the template tells the model it may call one or more functions and renders every call in a turn. Structured output 0, Neither the card nor the files mention JSON or JSON Schema output, beyond json as a tool-call format. Reasoning control 10, the card documents enable_thinking, on by default and False per request, and preserve_thinking for earlier turns, and names the nanbeige reasoning parser, which returns the reasoning apart from the answer. Context 10, 262,144 tokens, which the card says the model supports and config.json gives as max_position_embeddings. 20 + 25 + 20 + 0 + 10 + 10 = 85.

OTel-2.0-LLM-31B-IT 60

Chat template 20, chat_template.jinja in the repository, 19,300 characters, the publisher's own. Tool-calling format 10, only the template says how tools are declared and how calls come back. Trained tool calling 20, the template loops over tool_calls, so it renders several calls in one turn, and the card says the training mix included general-purpose agentic tool-calling examples, while none were specific to telecoms. Structured output 0, the card doesn't document JSON or JSON Schema output. Reasoning control 0, the card says the model wasn't trained to produce or consume Gemma's thought channel and answers directly, so there is no thinking to switch or budget. Context 10, 262,144 positions in the text config, the base's native context. 20 + 10 + 20 + 0 + 0 + 10 = 60.