# SPEV (slim)

> SPEV is Sprag's open-weight decision model, released on 14 October 2026. It answers predicate, choice and score questions about text, audio or an image with a probability for each answer, through Sprag's hosted API or Apache-2.0 weights.

- Full: https://www.anchorterminal.com/tools/sprag-spev.md (~7,100 tokens) · this version ~1,530 tokens · JSON https://www.anchorterminal.com/tools/sprag-spev.json · canonical https://www.anchorterminal.com/tools/sprag-spev
- Index: https://www.anchorterminal.com/llms.txt · API: https://www.anchorterminal.com/api/v1/index.json · Updated: 2026-10-15
- Licence: CC BY 4.0, credit Anchor Terminal (anchorterminal.com) and link this page when you pass anything on

**D · 51.1/100 · rank #819 of 1094 · #12 in Decision models · not agent-ready · confidence medium**


Assessment: SPEV answers typed questions over text, audio or an image in one call, with a public OpenAPI schema and Apache-2.0 weights on Hugging Face. It is a day old, has an 8,192-token context, no status page or changelog, and Sprag's terms let it train on API content unless the customer opts out by email.

## Facts

- Kind: Model API · vendor: Sprag · category: Decision models · legal entity: Sprag AI, Inc. · provenance 51/100
- Endpoint: `https://api.sprag.ai/v1/decisions` (HTTP)
- Auth: API key · pricing: Pay per use · x402: no · licence: Apache-2.0 (weights and inference code, following the Qwen3-Omni base). Training code and data aren't published
- Probe metrics: not measured yet (probes haven't run)
- Surface listed: The hosted `POST /v1/decisions` endpoint with model `spev`, read from the docs, the OpenAPI document and `GET /v1/models` on 15 October 2026. The open weights were read on Hugging Face, not downloaded or run
- Question types: predicate (probability a condition holds), choice (2 to 255 string or boolean values) and score (2 to 10 ordered levels). Up to 200 questions a request in the OpenAPI schema
- Evidence: Text or user messages, one image (data URL or public HTTPS URL), one audio clip as base64 `input_audio`. The OpenAPI schema and `GET /v1/models` also list `input_video`, which the Decisions docs do not describe
- Context: 8,192 tokens on the hosted model and `max_tokens` 8192 in the weights' config
- Model: 30B, Qwen3-Omni-30B-A3B Thinker with a rank-32 LoRA adapter and a choice-scoring head. Fine-tuned in part on the AMI Meeting Corpus (CC BY 4.0) per the third-party notices
- Weights: Apache-2.0 on Hugging Face, ungated, about 63 GB in 13 shards plus a 56 MB decision head. Loads with `trust_remote_code=True`. Training code and data are not published
- Rate limits: 60 requests a minute per organisation by default. Keyless requests are limited to 20 a minute and 300 a day per IP address
- Trains on API data: Yes by default under the terms. Opt-out by emailing privacy@sprag.ai, applied to later Content only
- Streaming and batch: No streaming mode. No batch endpoint documented
- Age: One day old on 15 October 2026. Sprag AI, Inc. has run its hosted API since March 2026
- Prices: SPEV input text or image $0.09 per 1M tokens; SPEV input audio $2 per 1M tokens

- Scores: Reliability 45, Performance pending, Schema & documentation 81, Agent ergonomics 76, Security & auth 33, Payments & pricing 30, Task success pending, Maintenance & community 49, Transparency & trust 54 · negative events -2 · total over the 7 assessed categories
- Why: Reliability, Read with the hosted lines, for the API an agent calls. · Schema & documentation, OpenAPI 3.1 at api.sprag.ai/openapi.json with typed request schemas for all three question types. · Agent ergonomics, Read as an API an agent calls for a decision. · Security & auth, Read for a model API. · Payments & pricing, No x402, MPP or L402 (0). · Maintenance & community, Read for a model. · Transparency & trust, Weights and inference code under Apache-2.0 with third-party notices for the Qwen base and the AMI corpus.
- Sources: 17, open questions: 7, both in the full twin
- Capabilities: inference.decision
- JSON: https://www.anchorterminal.com/api/v1/tools/sprag-spev.json
- Verify (for the vendor): the badge `https://www.anchorterminal.com/badges/sprag-spev.svg` or a link to https://www.anchorterminal.com/tools/sprag-spev from a page on sprag.ai or one of its subdomains, then `POST https://www.anchorterminal.com/api/v1/verify` `{"slug", "url"}` or `verify_listing` at /mcp; re-checked weekly, no effect on the grade. Snippets in the full twin.

## Before you call it

1. Send `model: "spev"` to `POST https://api.sprag.ai/v1/decisions` with a bearer key. `/v1/systemone` returns 410
2. Ask every question about one piece of evidence in one request. Answers come back in question order
3. Send audio as base64 in `input_audio` with `format`, not as a data URL, and at most one image across all messages
4. Keep evidence and questions under 8,192 tokens. Over the limit returns 400 `model_limit_exceeded`
5. Back off on 429 and 503 for `Retry-After` seconds. Limits are 60 requests a minute per organisation by default
6. Handle an answer of type `refusal`, which carries no probabilities

## Connect

```bash
curl https://api.sprag.ai/v1/decisions \
  -H "Content-Type: application/json" -H "Authorization: Bearer $SPRAG_API_KEY" \
  -d '{"model":"spev","input":"I was charged twice. Please refund the duplicate payment.","questions":[{"name":"department","type":"choice","instructions":"Which team should handle this?","choices":[{"value":"billing"},{"value":"technical"}]},{"name":"refund","type":"predicate","instructions":"Is the customer asking for a refund?"}]}'
```

## Similar tools

| Tool | Grade | Score | Shared capabilities | Slim |
| --- | --- | --- | --- | --- |
| OpenAI Decisions API | BB | 71.5 | inference.decision | https://www.anchorterminal.com/tools/openai-decisions-api.min.md |
| Decider | B | 69.5 | inference.decision | https://www.anchorterminal.com/tools/decider.min.md |
| Drex 1.5 | B | 69.5 | inference.decision | https://www.anchorterminal.com/tools/nace-drex.min.md |
| Laya | B | 69.2 | inference.decision | https://www.anchorterminal.com/tools/convai-laya.min.md |
| Kev | B | 67.4 | inference.decision | https://www.anchorterminal.com/tools/jaredpalmer-kev.min.md |


## Panel reviews (0, desk reviews from public material, no calls made)