Deem 0.8B v1
system-one-open · by LibertAI · Code and weights marked open
Base model: Qwen/Qwen3.5-0.8Bsourcev1.5 roster addendum A4
Not independently recorded
JevBench v1.5.4 score
2.138
Option A: rank #75 of 106 ranked systems.
The three option scores and ranks are published independently; the headline is Option A.
| Option | Score | Rank |
|---|---|---|
| A · headline | 2.138 | #75 of 106 |
| B | 1.664 | #75 of 106 |
| C | 1.485 | #75 of 106 |
Published axes
- intelligence
- 13.0
- calibration
- 38.9
- speed
- 84.3
- cost
- 80.3
Bands use the published 0–100 axis values; the marked reference is Jev 1.13.0 when that axis is available.
Run and cost evidence
- Run status
- complete · 1,624 decisions · 0 missing
- Cost
- $0.0045 per 1,000 decisions · estimate · ESTIMATE: documented hosted-model estimate. Frozen market reference deepinfra:Qwen/Qwen3.5-0.8B.
- Median latency
- 0.501 seconds, adjusted · x2 + 0.15 s (assumption, not measured)
- Endpoint condition
- evaluator-owned H100 GPU
- Published source
- https://huggingface.co/LibertAIDAI/deem-0.8-v1
Model and serving disclosure
LibertAIDAI/deem-0.8-v1 @ 8cbabbb2c4a7ef13c6b43f0ef3ae4157983c6d21; root model.safetensors SHA-256 80438125177c0681e855158d46720cbbd4ff07ea620cd55b0f456aafbfcf778d. Author Python GPU server, default DEEM_N_ORDERS=1, raw softmax (temperature 1.0). A four-field shim translates the wire format. The submitted interface accepts choice labels without descriptions and noul instructions without criteria; score receives full level descriptions. The author Rust/AVX-512 latency claim was not measured.
Values come from the public v1.5.4 aggregate. Scores and ranks may change in a later release.
Read the full leaderboard, the v1.5.4 release page, and the published method.