Nanbeige4.2-3B vs Nemotron 3 Ultra 550B A55B

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.

Nemotron 3 Ultra 550B A55B 55

Memory earns 0, because the smallest 4-bit build, NVIDIA's 352 GB NVFP4, needs about 423 GB with the fifth added, above the 256 GB band, and no runtime library has a local build. Runtimes earn 20 from released vLLM, SGLang, llama.cpp and Transformers, with Ollama's cloud-only entry not counted. Builds earn 10 for the single NVFP4 build. Terms earn 10, because OpenMDW-1.1 is permissive with no threshold or use policy but isn't an OSI licence. Download earns 10, with no gate or account. Hosted earns 5, since NVIDIA's own API serves it and two providers host it.

Nanbeige4.2-3B 55

File safety 15, safetensors only, no .bin, .pt or .pth files (25). Provenance 15, the files sit in the Nanbeige organisation (Nanbeige LLM Lab), which the lab's arXiv report for this model links, so the publisher's own account, although the Hub hasn't verified it (10). Safety evaluation 0, no safety results on the card. Agent guidance 5, General cautions: the limitations section says outputs may be inaccurate, biased or harmful. Licence risk 15, Apache 2.0 has no use policy, no threshold and no clause letting the licensor change or revoke it. Security programme 5, BOSS Zhipin runs a security response centre for reporting flaws in its products, as read for the Nanbeige4.3-33B listing (5), though nothing links it to the model. 15 + 15 + 0 + 5 + 15 + 5 = 55.

Nemotron 3 Ultra 550B A55B 70

File safety earns 15, safetensors only but with 10 off because the documented commands pass trust-remote-code and the repository ships Python files. Provenance earns 15, files in NVIDIA's linked organisation with a hash for every file. Safety evaluation earns 0, since NVIDIA describes safety training but publishes no results. Agent guidance earns 15, as the technical report documents training the model to ignore instructions injected through tool output, beside the subcard's warnings and safeguards. Licence risk keeps all 15 under OpenMDW-1.1. The security programme earns 10 for the reporting route on the card and NVIDIA's public bulletins.

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.

Nemotron 3 Ultra 550B A55B 90

The chat template earns 20, a Jinja file in the repository. The tool-calling format earns 25, since the card names the qwen3_coder parser and shows a tool call declared and read back. Trained tool calling earns 20, with tool-use training and several calls per turn in the template. Structured output earns 5, claimed from training with no instructions. Reasoning control earns 10, thinking off or medium effort per request with a runtime parser that returns the reasoning apart. Context earns 10 for up to 1M tokens.