# Mistral Large 4 vs Nanbeige4.2-3B (slim)

> Nanbeige4.2-3B scores 61 (C) to Mistral Large 4's 54.6 (C) for open-weight models you run yourself. Prices, MCP, x402, uptime and agent notes side by side.

- Full: https://www.anchorterminal.com/compare/mistral-large-4-vs-nanbeige-4-2-3b.md (~9,950 tokens) · this version ~780 tokens · JSON https://www.anchorterminal.com/compare/mistral-large-4-vs-nanbeige-4-2-3b.json · canonical https://www.anchorterminal.com/compare/mistral-large-4-vs-nanbeige-4-2-3b
- Index: https://www.anchorterminal.com/llms.txt · API: https://www.anchorterminal.com/api/v1/index.json · MCP: https://www.anchorterminal.com/mcp · Updated: 2026-11-10
- Licence: CC BY 4.0, credit Anchor Terminal and link this page (https://www.anchorterminal.com/terms/#data)

Nanbeige4.2-3B scores 61 (C) on agent readiness against Mistral Large 4's 54.6 (C), and leads in 2 of 4 scored categories. Mistral Large 4 leads on risk and agent interface. Both do open-weight models you run yourself.

- Mistral Large 4: C 54.6, rank #128 of 160 open-weight models · https://www.anchorterminal.com/tools/mistral-large-4.min.md
- Nanbeige4.2-3B: C 61, rank #78 of 160 open-weight models · https://www.anchorterminal.com/tools/nanbeige-4-2-3b.min.md
- Mistral Large 4, good for Operators with an 8-GPU Blackwell or H200 node who want a frontier-sized open model for coding agents and defensive security, with a hosted twin on Mistral's API. Ahead on Risk, 75 against 55; Agent interface, 95 against 85.
- Nanbeige4.2-3B, good for Local agents that need tool calling and a long context on a laptop or a small GPU Ahead on Reach, 85 against 50; Documentation and upkeep, 72 against 64.

| Category | Mistral Large 4 | Nanbeige4.2-3B |
| --- | --- | --- |
| Capability (35%) | pending | pending |
| Reach (25%) | 50 | 85 |
| Risk (20%) | 75 | 55 |
| Agent interface (10%) | 95 | 85 |
| Documentation and upkeep (10%) | 64 | 72 |

| Fact (where they differ) | Mistral Large 4 | Nanbeige4.2-3B |
| --- | --- | --- |
| Vendor | Mistral AI | Nanbeige LLM Lab (BOSS Zhipin) |
| Licence | Mistral Ecosystem Open-Weight License (modified MIT) | Apache-2.0 (Hub metadata only; the repository has no LICENSE file) |
| Last release | 2026-10-29 | 2026-09-11 |

Why the edge:
- Reach, Nanbeige4.2-3B +35: Mistral Large 4 50 (Memory scores 0, since the smallest 4-bit build, Mistral's NVFP4, is 610 GB, about 732 GB with a fifth added, far above 256 GB.); 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.)
- Risk, Mistral Large 4 +20: Mistral Large 4 75 (File safety scores 25, safetensors only and loaded by Transformers' own code with a Jinja template.); Nanbeige4.2-3B 55 (File safety 15, safetensors only, no .bin, .pt or .pth files (25).)
- Agent interface, Mistral Large 4 +10: Mistral Large 4 95 (The chat template is in the repository (20).); 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 switche…)
- Documentation and upkeep, Nanbeige4.2-3B +8: Mistral Large 4 64 (The card covers intended use, architecture, parameters, context and languages, and misses limitations and a cut-off (14).); Nanbeige4.2-3B 72 (Model card 17, Intended use, limitations, the architecture, the parameters (3B outside the embeddings), the context and the languages (in the metadata) are on t…)