OpenJev GGUF

✓ Verified by Badgr · llama.cpp · Serve

Last verified 2026-10-10

badgr serve --runtime llama.cpp --hf-repo openjev/openjev-GGUF --hf-file OpenJev-Q4_K_M.gguf --gpu RTX_4090 --funding paid --max-cost 1.5 --env LLAMA_ARG_N_GPU_LAYERS=999 --env LLAMA_ARG_CTX_SIZE=16384 --env LLAMA_ARG_N_PARALLEL=2

GPU

NVIDIA GeForce RTX 4090Reported by the provider, not confirmed inside the container

Time to verified

362s

Deployment

dep-7780d1fffb

• ✓ llama-server loaded the 16.5 GB Q4_K_M file with the model card's settings (-ngl 999, -c 16384, -np 2)
• ✓ The model card's own decision prompt returned the correct letter
• ✓ Loop AI's helper/shim.py, unmodified with default settings, answered 3 of 3 choice questions correctly at 0.998-0.999 confidence
• ✓ A combined choice, yes/no and score request returned in 1.5 s
• ✓ final_state = VERIFIED
• ✓ Teardown completed

Weights are licensed CC BY-NC 4.0 (research and non-commercial use); a commercial licence comes from Loop AI. The shim ran on a separate machine against the endpoint's OpenAI-compatible API. Three hand-written questions are a smoke test, not an accuracy measurement; the model card reports 82.8% on 1,789 rows for this file.

final_state=VERIFIED failure_class=None providers_tried=1 spend_usd=0.0443 elapsed=362.4s
shim: complete-purchase click_place_order p=0.9994 | enter-discount-code click_coupon_field p=0.9994 | search-flights type_destination p=0.9984

Run from Badgr's local development environment. Documents that this deployment path works end-to-end for this model.