llms.txt
# Decisions API
> Decisions API brings decision models together in one playground and API. Classify, score, and route text with structured answers. Start with 100 credits.
## Pages
- [Muse Wonder vs Gemini 4 Argon](https://decisions-api.dev/muse-wonder-vs-gemini-4-argon): Public benchmark and pricing comparison, reviewed October 14, 2026. Artificial Analysis Intelligence Index v4.3.2: ordinary Muse Wonder (Max) 52 vs Gemini 4 Argon (High) 53, with ten component results. Lab-reported results are separate. Compares current upstream prices, undiscounted credit-pricing bases and published platform rates, with primary sources and links to both shared playgrounds.
- [Gemini 4 Argon](https://decisions-api.dev/model/gemini-4-argon): Google multimodal reasoning model via OpenRouter. Text, image, video, audio and PDF inputs; text output; 1,048,576-token context. Shares the Muse Wonder playground at /multimodal?model=google%2Fgemini-4-argon and /v1/chat/completions API. Platform output limit: 8,192 tokens. Credits use undiscounted list prices ($4 input, $20 output, $0.20 cache per million tokens), with the existing 30% margin rule, independent of the current 50% promotion.
- [Classify-1](https://decisions-api.dev/model/classify-1): ScaleDown decision model through OpenRouter using /v1/systemone. Choice, Score and Noul; text and JSON (serialized to text), no image or PDF input. 1,000,000-token provider context; platform: 8 questions, 32 KiB text. 720 credits/million actual input tokens based on the undiscounted $0.05/M list price, independent of the current $0.04/M promotion; minimum 1 credit per successful request; output free.
- [Solar Decide Flash](https://decisions-api.dev/model/solar-decide-flash): Upstage decision model through OpenRouter using the shared Jev /v1/systemone API. Text and JSON; Choice (2–26 options), Score and Noul; no image input. 720 credits/million input tokens, minimum 1 per successful request; output free. Provider context: 524,288 tokens; platform: 8 questions and 32 KiB text plus questions.
- [GPT-6 Luna Decisions](https://decisions-api.dev/model/gpt-6-luna-decisions): OpenAI decision model through OpenRouter using the shared Jev /v1/systemone API. Text, JSON and images; Choice, Score and Noul. 1,500 credits/million input tokens, minimum 1 per successful request; output free. Provider context: 1,050,000 tokens; platform: 8 questions and 32 KiB text.
- [Perplexity Decider V1 27B](https://decisions-api.dev/model/pplx-decider-v1-27b): Jev-compatible Choice, Score and Noul through OpenRouter. Text and JSON state, 262,144-token provider context. 600 credits/million input tokens, minimum 1; output free.
- [Clef Flash](https://decisions-api.dev/model/clef-flash): Cloudflare decision model through OpenRouter. Choice, Score and Noul. Provider currently reads about the first 2K text-state tokens; this service caps serialized state at 2,000 UTF-8 bytes. 1,300 credits/million input tokens, minimum 1; output free.
- [Clef](https://decisions-api.dev/model/clef): Cloudflare decision model through OpenRouter. Choice, Score and Noul. Provider currently reads about the first 2K text-state tokens; this service caps serialized state at 2,000 UTF-8 bytes. 3,500 credits/million input tokens, minimum 1; output free.
- [Laya](https://decisions-api.dev/model/laya): Convai Innovations decision model through Vercel AI Gateway. Choice, Score and Noul; 8K context; 1 credit per successful web or API request.
- [Span-01](https://decisions-api.dev/model/span-01): Noul-only behavior evaluation. Web and API cost 300 credits/million input tokens, minimum 1.
- [Span-01 Lite](https://decisions-api.dev/model/span-01-lite): Lightweight behavior evaluation. Web and API cost 1 credit per successful request.
- [Home](https://decisions-api.dev): Landing page
- [Pricing](https://decisions-api.dev/pricing): Pricing plans
- [OpenAI Decisions API vs Jev](https://decisions-api.dev/openai-decisions-api-vs-jev): Compare Jev with GPT-6 Luna Decisions, now integrated through OpenRouter using the shared Jev API. Text and images, provider and platform pricing, native protocol differences, and verified limits. Reviewed October 7, 2026.
- [Kev 4B](https://decisions-api.dev/model/kev-4b): Credit-based playground and decision API; same input-token credit rate as Jev. New accounts receive 100 credits.
- [Solar Decide](https://decisions-api.dev/model/solar-decide): Upstage structured decision model with Jev-compatible System One input and output, available in the Jev playground and API.
