GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
caiovicentino1
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Model Details
Model Provider
caiovicentino1
Model Tree
caiovicentino1/Eikos-4B
Input Modalities
Output Modalities
Supported Functionality
Requires vLLM ≥ 0.30.0. Older builds return wrong answers when several long requests are batched together on this hybrid (Gated DeltaNet) architecture.
bash
hf download caiovicentino1/Eikos-4B-FP8 --local-dir Eikos-4B-FP8bash Eikos-4B-FP8/serve_vllm.sh $PWD/Eikos-4B-FP8 8001 # vLLM engine: letter readout + hybrid prefix cachepython Eikos-4B-FP8/serve.py --model $PWD/Eikos-4B-FP8 --vllm-url http://127.0.0.1:8001 --port 8000 # HTTP API on :8000The HTTP API, agent sessions and the question types are the same as for Eikos-4B.
Same 7,371 items for both builds (7 suites, never used in training), vLLM 0.30 with batching and prefix cache on. The release gate was fixed before looking at results: accuracy within 1 point of bf16, ECE within 0.01, and at least 97% of answers unchanged.
| Eikos-4B (bf16) | Eikos-4B-FP8 | |
|---|---|---|
| Size | 9.3 GB | 5.8 GB |
| JevBench public — original / hard | 91.7 / 73.9 | 90.3 / 73.0 |
| DecisionBench — medium / hard | 77.1 / 66.6 | 78.5 / 66.6 |
| General battery (9 tasks) | 75.7 | 75.8 |
| Finance (CUAD, sentiment, FinQA-judge) | 74.7 | 75.2 |
| Trade rules — seen / unseen | 74.5 / 76.0 | 74.7 / 76.0 |
| Compositional rules — same type / new domain / rulebooks | 96.0 / 91.5 / 91.7 | 95.6 / 91.8 / 91.3 |
| ECE (lower is better) | 0.033 | 0.031 |
| ≥0.90 confidence: decides / error | 34.7% / 2.5% | 34.8% / 2.5% |
| Same answer as bf16 (all / confident ≥0.9) | — | 97.3% / 100.0% |
It passes our release gate.
MIT for our contributions (LICENSE). The base model, Qwen3.5-4B, is Apache-2.0 (LICENSE-Qwen);
attributions are in NOTICE. Not legal, tax or investment advice.
hf download caiovicentino1/Eikos-4B-FP8 --local-dir Eikos-4B-FP8bash Eikos-4B-FP8/serve_vllm.sh $PWD/Eikos-4B-FP8 8001 # vLLM engine: letter readout + hybrid prefix cachepython Eikos-4B-FP8/serve.py --model $PWD/Eikos-4B-FP8 --vllm-url http://127.0.0.1:8001 --port 8000 # HTTP API on :8000