Vela 2.0 0.8B
Open Foundation Routing Models
Routing decisions. Safety checks. Precise text spans.
The compact hybrid member of Vela 2.0: a 756M-parameter model for multilingual routing, safety checks and span-level decisions through one interface.
Define options, labels and rubrics at request time. Ask multiple named questions about a request, context and answer, and receive structured decisions with the text spans that support your workflow.
- Compact deployment. About 3 GB of GPU memory for FP32 parameters, with a 16,384-token input limit.
- Routing and safety together. Use one request for routing, prompt-attack checks, PII and unsupported-claim detection.
- Decisions at span resolution. Return precise text locations for trained router labels and open-label extraction, with probabilities and character offsets.
| Specification |
Value |
| Reported parameters |
756M |
| Backbone |
24-layer Qwen3.5 hybrid: Gated DeltaNet and gated GQA |
| Input limit |
16,384 tokens |
| Output |
Choice, Yes/no (Noul), Score, Span and Set; no text generation |
| Evaluated precision |
FP32 parameters; bf16 backbone autocast on GPU; FP32 heads |
| GPU parameter memory |
About 3 GB in FP32 |
| Licence |
Apache-2.0 |
Quickstart
Install the dependencies, load the model and ask for routing and spans in one request:
pip install torch "transformers>=5.17" safetensors tokenizers numpy
pip install flash-linear-attention # GPU: the Gated-DeltaNet kernel used in evaluation (optional, much faster)
from transformers import AutoModel
m = AutoModel.from_pretrained("vllm-sr/Vela-2.0-0.8B", trust_remote_code=True)
m = m.to("cuda") # parameters load in FP32; on GPU the backbone runs under bf16 autocast (heads FP32)
PII_LABELS = m.vela2_engine.cal["pii_schema"]["labels"] # the 17 trained PII types
result = m.system_one(
state={"request": "Hi, I'm Tom Baker ([email protected]). What is the maximum daily dose of paracetamol for an adult?",
"source": "For adults, the maximum dose of paracetamol is 4 grams in 24 hours, taken as 500 mg to 1 g every 4 to 6 hours.",
"answer": "Adults can take up to 6 grams of paracetamol in 24 hours, in doses of 500 mg to 1 g every 4 to 6 hours."},
questions={
"pii": {"type": "span", "instructions": "Which spans are personal information?", "criteria": PII_LABELS,
"over": "request"},
"halu": {"type": "span", "instructions": "Which spans of the answer are not supported by the context?",
"criteria": {"unsupported": "a claim not supported by the context"}}, # over the answer by default
"domain": {"type": "choice", "instructions": "Which subject area is this request about?", "over": "request",
"criteria": {"health": "medicine, clinical practice, nutrition, ageing or sexual health",
"math": "arithmetic, algebra, geometry, statistics or other mathematics",
"other": "a subject that fits none of the listed areas"}},
})
Recorded response excerpt from this release; the full response includes usage and thresholds:
{
"answers": {
"domain": {
"type": "choice",
"choice": "health",
"confidence": 0.929,
"probabilities": {
"health": 0.962,
"math": 0.005,
"other": 0.033
}
}
},
"spans": {
"pii": [
{
"label": "PERSON",
"start": 8,
"end": 17,
"text": "Tom Baker",
"probability": 0.984
},
{
"label": "EMAIL_ADDRESS",
"start": 19,
"end": 40,
"text": "[email protected]",
"probability": 0.985
}
],
"halu": [
{
"label": "unsupported",
"start": 19,
"end": 29,
"text": "to 6 grams",
"probability": 0.778
}
]
},
"span_heads": {
"pii": "router",
"halu": "router"
}
}
GPU with bf16 autocast is the evaluated setting. FP32 on CPU or GPU is also supported; fp16 is not. Parameter memory above excludes runtime allocations. See precision and execution.
Questions and outputs
| Question |
Use it for |
Output |
| Choice |
Route a request or select one of 2–255 supplied options. |
Selected key and distribution |
| Yes/no (Noul) |
Check a condition against the state. |
P(yes) |
| Score |
Rate against 2–10 ordered levels. |
Expected level and distribution |
| Span |
Locate personal information, unsupported claims, entities, relations or evidence. |
Text spans, character offsets and probabilities |
| Set |
Select any number of supplied labels. |
Selected labels and per-label probabilities |
Two span heads, one interface. The router span head handles trained PII, hallucination and toxic labels. The broad span head handles open-label extraction. The engine selects a head for each span question; you can also set "head": "router" or "head": "broad". See span-head selection.
Span offsets refer to Unicode code points in the selected text. Span and Set answers also provide Yes/no views in the SystemOne response. Output semantics and thresholds explain these fields.
Results
Selected results for this checkpoint. Full evaluation includes every benchmark, family comparison and scoring protocol.
| Benchmark |
Result |
Metric and mode |
| Safety, 14 public sets |
0.875 |
Mean AUC; trained task families |
| Prompt attacks, unseen families |
0.940 |
AUC |
| PII, short texts |
0.984 |
F1; shipped calibration |
| ACL-Verbatim evidence |
23.6 |
Word-F1; broad span head, zero-shot |
- Evaluation: checkpoint selection, temperatures and thresholds used dev data. Safety results cover trained task families; they are not zero-shot comparisons.
- Scoring: PII uses shipped calibration; ACL-Verbatim was held out of training. Benchmark protocols.
Architecture
This model has its own 24-layer hybrid backbone, with Gated DeltaNet and gated GQA blocks, a candidate head, and separate router/broad span-head weights. Editable SVG.
Operator and readout diagrams
**Gated DeltaNet**
Causal QKV convolution feeds the gated-delta update, followed by per-head normalization and output gating. [Editable SVG](assets/architecture/07-gated-deltanet.svg).
**Gated GQA and SwiGLU**
Gated GQA uses partial RoPE on Q/K and a sigmoid output gate; SwiGLU uses separate up and gate projections. [Editable SVG](assets/architecture/08-gated-gqa-swiglu.svg).
**Candidate readout**
CandidateHead combines scaled bilinear and additive MLP scores over runtime options, then applies task-specific calibration. [Editable SVG](assets/architecture/09-decoder-candidate-head.svg).
**Router and broad span heads**
Router and broad span heads have separate weights, with one selected for each span question. Overlapping word logits are combined before decoding. [Editable SVG](assets/architecture/10-decoder-span-heads.svg).
**State prefix and question isolation**
For a fixed rendered state, each question block sees the same state prefix and its own causal tokens. Additional span questions use separate sequences. [Editable SVG](assets/architecture/11-state-prefix-and-question-isolation.svg).
Choice, Yes/no, Score and Set share a candidate readout. Router and broad span heads use the same word-by-label topology with separate weights. The hybrid backbone reuses the state prefix within each rendered sequence; execution details are in USAGE.md.
Reference
| Document |
Contents |
| USAGE.md |
Full examples and recorded responses, typed parts, head selection, calibration, SDK and HTTP serving |
| EVALUATION.md |
Complete family results, research versus shipped PII modes, benchmark protocols and evaluation disclosures |
| TRAINING.md |
Training stages, data mixtures, provenance, licences and release files |
| PARITY.md |
Export parity against the research scorer |
Credit and citation
Vela 2.0 is led by KR Labs and vLLM Semantic Router.
Read the Vela 2.0 technical overview.
@misc{vela2_08b_2026,
title = {Vela 2.0: Towards Open Foundation Routing Models},
author = {{KR Labs} and {vLLM Semantic Router}},
year = {2026},
note = {Blog post. Model: vllm-sr/Vela-2.0-0.8B},
howpublished = {\url{https://vllm-sr.ai/blog/vela-2-0-open-foundation-routing-models/}}
}
Licence
Apache-2.0 for this model's weights, code and documentation. It is derived from Decision-2.0-Eos-0.8B (Apache-2.0), itself built on Qwen3.5-0.8B (Apache-2.0); their licences and notices are passed on in LICENSE, NOTICE, ATTRIBUTIONS.md and LICENSES/. Changes against Decision-2.0-Eos-0.8B: MODIFICATIONS.md. Training data keep their own licences, some of them CC-BY-SA share-alike.
