elephant-1 vs JEV-27B-VL

elephant-1

A small cousin for routing, triage and judging, answering several typed questions about one state with probabilities in one prefill. It fits a 24 GB GPU in BF16 but runs only through a patched third-party runtime. The risks are a licence switched from proprietary to MIT on release day with no licence file, and a thin card.

JEV-27B-VL

A cousin for typed decisions with probabilities over text and images, answering in one pass with optional thinking. It needs about 67 GB in BF16 on vLLM, with no AutoTrust 4-bit build or hosted API. The main risks are labels distilled from TypeSafe's Jev against its customer terms and serving through a repository script on a vLLM development build.

elephant-1

1. Start on a CUDA GPU in pytorch/pytorch:2.8.0-cuda12.8-cudnn9-devel with git clone https://huggingface.co/baselabs/elephant-1 && cd elephant-1 && bash setup.sh /tmp/w, which clones OneJev at 81ce62f, applies serve.patch and installs the Decision Index kit
2. Then qev serve --model . --port 8000 is OneJev's documented server. Base Labs doesn't document it for this checkpoint, so check the answers before relying on them
3. Don't load it with stock vLLM or Transformers. They ignore the block-bidirectional attention in config.json's modeling_overrides and give probabilities the model wasn't trained for
4. Keep each request under 16,384 tokens, the trained length, or the engine refuses it
5. Group the questions about one state in one request, since the state is read once and shared
6. Without hardware, Baseten's /v1/systemone serves it as decision-models/baseten-elephant-9b-latest with a Baseten key, free until the end of the launch week

JEV-27B-VL

1. Quickest start is hf download autotrust/JEV-27B-VL then bash JEV-27B-VL/serve.sh, which serves POST /v1/decide on port 8000, but it runs the repository's serve_decide.py on a September 2026 vLLM development build
2. On a released vLLM use plain vllm serve with --enable-lora --lora-modules jev-decision=JEV-27B-VL/adapter_vllm and do the card's client-side arithmetic with decision_head.json and calibration.json
3. Keep --max-num-seqs 8, the card says larger batches return wrong System 1 probabilities on vLLM's multimodal LoRA path
4. When calling /v1/completions or /v1/chat/completions yourself pass top_k 0 and top_p 1.0, or generation_config.json's top_k 20 and top_p 0.95 truncate the probabilities
5. Choices beyond 16 options use labels the head never saw in training, and the response's adaptation field says wide-labels
6. Downscale images to about 448 px to keep vision tokens and latency down
7. Treat it as AutoTrust's model, not TypeSafe's, and its request shape differs from TypeSafe's /v1/systemone