Head to head · Decision models · October 2026 research run
Celeris-1 Decision vs Decision 3.0
Decision 3.0 scores 63.1 (B) on agent readiness against Celeris-1 Decision's 49.3 (D), and leads in 5 of 7 scored categories. Celeris-1 Decision leads on agent ergonomics and transparency & trust. Both do decision models. Celeris-1 Decision is cheaper for decision models, $0 against $0 per 1M tokens.
Best decision models for AI agents · All 105 decisions comparisons
Which one, for what
Best for Multimodal classification, routing and bounded decisions with probabilities and optional explanations
Ahead on
- Agent ergonomics, 80 against 72
- Transparency & trust, 53 against 45
Also in its favour
- Cheaper for decision models, $0 against $0 per 1M tokens
- A hosted endpoint, with nothing to install
Watch for
Early-access terms and capacity-dependent workspace activation
Best for Self-hosted typed decisions over screenshots, documents and short video as well as text, where open weights matter.
Ahead on
- Reliability, 58 against 40
- Schema & documentation, 70 against 60
- Security & auth, 56 against 45
- Payments & pricing, 60 against 20
- Maintenance & community, 82 against 40
Also in its favour
- No key needed to call it
- Open source
Watch for
Released on 10 October 2026, with five tagged releases of each model by 11 October
Score by category
| Category | Weight this run | Celeris-1 Decision | Decision 3.0 | Edge |
|---|---|---|---|---|
| Reliability | 16%20 | 40 | 58 | Decision 3.0 +18 |
| Performance | 10%pending | pending | pending | not scored in this run |
| Schema & documentation | 13%16.2 | 60 | 70 | Decision 3.0 +10 |
| Agent ergonomics | 13%16.2 | 80 | 72 | Celeris-1 Decision +8 |
| Security & auth | 14%17.5 | 45 | 56 | Decision 3.0 +11 |
| Payments & pricing | 10%12.5 | 20 | 60 | Decision 3.0 +40 |
| Task success | 10%pending | pending | pending | not scored in this run |
| Maintenance & community | 7%8.8 | 40 | 82 | Decision 3.0 +42 |
| Transparency & trust | 7%8.8 | 53 | 45 | Celeris-1 Decision +8 |
| Negative events | ≤15 | 0 | 0 | |
| Total | D 49.3/100 | B 63.1/100 |
Facts side by side
| Fact | Celeris-1 Decision | Decision 3.0 |
|---|---|---|
| Kind | Model API | Model API |
| Vendor | Celeris (Marqo Inc) | vLLM Semantic Router Team |
| Hosted endpoint | https://inference.celeris.ai/celeris-1-decision/v1/systemone | no (local only) |
| Transports | HTTP | HTTP |
| Auth | API key | None |
| Pricing | Pay per use | Free |
| Price for decision models | free | free |
| x402 | no | no |
| Licence | Proprietary hosted model under Celeris terms of service | Apache-2.0 for the weights and the bundled runtime, built on Apache-2.0 Qwen3.5 and Qwen3.8 bases |
| Read-only variant documented | no | no |
| llms.txt | yes | no |
| Last release | 2026-10-08 | 2026-10-11 |
| Terms last updated | 2026-09-08 | no document linked |
| Privacy policy last updated | 2026-07-23 | no document linked |
| Customer content may train models | not found in the text | |
| Terms restrict automated access | not found in the text | |
| Terms restrict benchmarking | yes | |
| Terms or service can change without notice | not found in the text | |
| Arbitration or class-action waiver | not found in the text |
Verdicts
Celeris-1 Decision
Typed probabilities, image inputs and optional explanations suit routing and classification inside an agent. Both System One and OpenAI Decisions request formats are documented. The service remains early access under its terms, activation can queue, and prepaid credit expires after 30 days. The published accuracy and latency figures are Celeris measurements, not Anchor Terminal tests.
Decision 3.0
Open Apache-2.0 decision models that read images and video as well as text and answer several typed questions in one call, with a probability per option and tagged Hub releases. They are two days old, need a GPU and trust_remote_code, drop Vela's set and span types, and publish no calibration figures.
Before you call either
Celeris-1 Decision
- Use the decision model path and matching model field. This model has no chat, Responses or models endpoint.
- Use at most 512 questions per request, or 64 with explanations. Both endpoints reject streaming.
- Preserve x_celeris in a custom Jev SDK response model when reading explanations. The default response type drops it.
- Retry 429 with Retry-After and backoff. Stop on 402 until workspace credit is replenished.
- Read usage.input_tokens for cost, including images and request overhead. Track credit expiry separately from token consumption.
Decision 3.0
- Pin a tag or commit with
revision=when loading from the Hub. Each model had five tagged releases on 10 and 11 October - Send images and videos to
d3_server.pyas base64 data URLs only. The server refuses paths and http URLs with 422 - Set
DECISION_API_KEYbefore binding the bundled server beyond 127.0.0.1. It is open otherwise - Turn on
permutation_average=Truewhen routing must not depend on option order. It is off by default - Expect one request at a time from the bundled server. Use the router's model runtime for queued serving
Questions
Which is better for AI agents, Celeris-1 Decision or Decision 3.0?
Decision 3.0 scores 63.1 (B) on agent readiness against Celeris-1 Decision's 49.3 (D), and leads in 5 of 7 scored categories. Celeris-1 Decision leads on agent ergonomics and transparency & trust.
Which is cheaper for decision models, Celeris-1 Decision or Decision 3.0?
Celeris-1 Decision, at free against free for Decision 3.0. These are the vendors' published prices for the job.
Do Celeris-1 Decision and Decision 3.0 need an API key?
Celeris-1 Decision needs an API key. Decision 3.0 needs no key.
Can an agent call Celeris-1 Decision and Decision 3.0 without installing anything?
Celeris-1 Decision has a hosted endpoint at https://inference.celeris.ai/celeris-1-decision/v1/systemone. No hosted endpoint is listed for Decision 3.0.
Are Celeris-1 Decision and Decision 3.0 open source?
No open-source release is listed for Celeris-1 Decision. Decision 3.0 is open source (Apache-2.0 for the weights and the bundled runtime, built on Apache-2.0 Qwen3.5 and Qwen3.8 bases).
Other comparisons with Celeris-1 Decision or Decision 3.0
- Celeris-1 Decision vs Clef
- Celeris-1 Decision vs Laya
- Celeris-1 Decision vs Decider
- Celeris-1 Decision vs GLiClass
- Celeris-1 Decision vs Kev
- Celeris-1 Decision vs Liquid d1
- Celeris-1 Decision vs Microsoft-Decision-1
- Celeris-1 Decision vs Drex 1.5
- Celeris-1 Decision vs OpenAI Decisions API
- Celeris-1 Decision vs pplx-decider
- Celeris-1 Decision vs Strands Decider 2B
- Celeris-1 Decision vs Jev
- Celeris-1 Decision vs Vela 2.0
- Clef vs Decision 3.0
- Laya vs Decision 3.0
- Decider vs Decision 3.0
- Decision 3.0 vs GLiClass
- Decision 3.0 vs Kev
- Decision 3.0 vs Liquid d1
- Decision 3.0 vs Microsoft-Decision-1
- Decision 3.0 vs Drex 1.5
- Decision 3.0 vs OpenAI Decisions API
- Decision 3.0 vs pplx-decider
- Decision 3.0 vs Strands Decider 2B
- Decision 3.0 vs Jev
- Decision 3.0 vs Vela 2.0
Machine-readable
- This page as Markdown
/compare/celeris-1-decision-vs-decision-3.md· slim.min.md· JSON.json(or sendAccept: text/markdown) - Each listing in full
/api/v1/tools/celeris-1-decision.json·/api/v1/tools/decision-3.json - From a terminal
anchor compare celeris-1-decision decision-3(the CLI) - Over MCP
compare_tools {"a": "celeris-1-decision", "b": "decision-3"}at/mcp, no key