OLMoE 1B/7B
可交互Validated on the recorded professional hardware profile.
7B total1B active per token
- 运行
- 可运行
- 验证
- 已验证
- 可用性
- 可交互
- 硬件
- Professional 硬件配置档
- 供应状态
- 已批准的配置档
- 证据来源
- AI Stays Local
- 发布者
- Allen Institute for AI
- 许可证
- Apache-2.0
- 来源
allenai/OLMoE-1B-7B-0125-Instruct- 固定的修订版本
b89a7c4bc24fb9e55ce2543c9458ce0ca5c4650e- 量化
- int8 merged experts (upstream conversion tool)
- 磁盘上的检查点
- 8 GB
- 测试过的上下文长度
- 2048 个 token
核查了什么Cited-document workflow validated: retrieval, citation of file and location, and refusal when the indexed documents do not contain the answer.
限制- Small active model: one billion parameters participate in a token. Instruction following is inconsistent on difficult prompts.
- No professional acceptance suite has been run, so no professional workflow claim is made.
- The local API answers stream=true with buffered transport, so the first token arrives with the whole answer.
来源信息Weights by Allen Institute for AI under Apache-2.0. Converted to int8 by the upstream Colibri conversion tool. Neither the weights nor the conversion are work of AI Stays Local.
测量、硬件与方法
Qwen3.5 122B-A10B
可交互Validated on the recorded professional hardware profile.
122B total10B active per token (256 experts, top-k 8, 48 layers)
- 运行
- 可运行
- 验证
- 已验证
- 可用性
- 可交互
- 硬件
- Professional 硬件配置档
- 供应状态
- 仅限私下验证
- 证据来源
- AI Stays Local
- 发布者
- Alibaba Qwen
- 许可证
- Apache-2.0
- 来源
Qwen/Qwen3.5-122B-A10B-GPTQ-Int4- 固定的修订版本
30cd92cba9707a9aba09d1e490ed4b66b78e9606- 量化
- GPTQ INT4, group size 128, symmetric; attention, shared experts, embeddings, lm_head and the vision tower are not quantised
- 磁盘上的检查点
- 79 GB
- 测试过的上下文长度
- 8192 个 token
核查了什么Thirteen of thirteen functional checks, six of six product API checks and the document gates pass, including citation and refusal.
限制- No professional acceptance suite has been run, so no professional workflow claim is made.
- Requires the machine to itself: peak unified memory 90.2 GiB.
- No distributable profile exists. Validation was private.
来源信息Weights and quantisation published by Alibaba Qwen under Apache-2.0. Served by vLLM (Apache-2.0). Neither is work of AI Stays Local; the validation is.
测量、硬件与方法
Qwen3.6 35B-A3B
可运行部分验证。取得证据的环境：professional 硬件配置档。
35B total3B active per token (256 experts, top-k 8)
- 运行
- 可生成
- 验证
- 部分验证
- 可用性
- 批处理
- 硬件
- Professional 硬件配置档
- 供应状态
- 不提供
- 证据来源
- AI Stays Local
- 发布者
- Alibaba Qwen (weights); third-party conversion
- 许可证
- Apache-2.0
- 来源
Kreuzzelg/qwen36-35b-a3b-colibri-i4-gs64- 固定的修订版本
c619aa594ad1e70af82168fb6b4878427896e21c- 量化
- int4 packed, group size 64
- 磁盘上的检查点
- 24 GB
- 测试过的上下文长度
- 2048 个 token
核查了什么Four of four instruction-following checks pass. Document gates were not run against this model.
限制- Below the interactive threshold at 3.68 tokens per second.
- The engine's OpenAI endpoint leaks a stop token into the content.
- Document-correctness gates were not run.
来源信息Third-party conversion of Qwen weights for the Colibri engine. Neither the weights nor the conversion are work of AI Stays Local. Revision c619aa594ad1e70af82168fb6b4878427896e21c established on 2026-09-21 by querying the Hugging Face model API for the repository and reading the sha field, which is the immutable commit of the default branch at that time. The licence field of the same response reported apache-2.0. The benchmark was taken before the revision was recorded, so the artifact relationship is asserted rather than proven byte for byte.
测量、硬件与方法
GLM-5.2 744B
研究演示Validated on the recorded professional hardware profile.
744B total~40B active per token (256 experts, top-k 8 plus one shared, across 75 of 78 layers)
- 运行
- 可运行
- 验证
- 已验证
- 可用性
- 仅限研究
- 硬件
- Professional 硬件配置档
- 供应状态
- 不提供
- 证据来源
- AI Stays Local
- 发布者
- Zhipu AI (weights); third-party conversion by mastouri
- 许可证
- MIT
- 来源
mastouri/GLM-5.2-colibri-int4-g64-with-int8-mtp- 固定的修订版本
6bbb01ed3e515a8730b694dfae73aadfd6774581- 量化
- int4 grouped scales gs=64, int8 embeddings and lm_head, int8 MTP head
- 磁盘上的检查点
- 429 GB
- 测试过的上下文长度
- 2048 个 token
核查了什么Ten of ten functional checks and eight of eight document checks pass, including resistance to a prompt injection planted in an indexed file.
限制- 0.17 tokens per second. A 200-token answer takes about 20 minutes.
- First token at 1018 seconds on a 311-token document prompt.
- Execution and validation are proven. Usability is research only.
来源信息Weights by Zhipu AI under MIT. Container converted by a third party (mastouri) under MIT, declaring zai-org/GLM-5.2 as its base model. Engine is upstream Colibri under Apache-2.0. None of the three is work of AI Stays Local; the validation is.
测量、硬件与方法