AI News Feed
Market watch
Le Monde publishes article on photography bicentenary and first person in a photographHeart Rate Variability in Wearables: Promise and UncertaintyDeepSeek Releases V4.1-Flash-1017 With 77.0 DeepSWE Score, Opens Search APIsMore Than 20 Jev Alternatives Emerge in Four Weeks as Decision-Model Competition GrowsStepFun Releases Step 5 Preview Open Weights for Agentic WorkAstronomers Identify First Known Green Star, a White Dwarf 326 Light-Years Away
Le Monde publishes article on photography bicentenary and first person in a photographHeart Rate Variability in Wearables: Promise and UncertaintyDeepSeek Releases V4.1-Flash-1017 With 77.0 DeepSWE Score, Opens Search APIsMore Than 20 Jev Alternatives Emerge in Four Weeks as Decision-Model Competition GrowsStepFun Releases Step 5 Preview Open Weights for Agentic WorkAstronomers Identify First Known Green Star, a White Dwarf 326 Light-Years Away
Products & Applications

More Than 20 Jev Alternatives Emerge in Four Weeks as Decision-Model Competition Grows

Jev drew at least 20 alternatives in four weeks, including Cloudflare, OpenAI and Perplexity, MarkTechPost reports.

AI News Feed · 2026-10-17 17:43

Jev is a decision model, not a chatbot. Users send a state such as a ticket or log line and typed questions in Choice, Score or Noul, the yes/no format. Jev scores every question in one parallel pass and generates no text. TypeSafe lists $0.042 per 1M input tokens with free output, a 64k-token context per request and text-only input. The weights are closed, and access runs through a waitlist. TypeSafe workflow evals put Jev at 67.8% agreement and 0.4 seconds per case. MarkTechPost said developers still cite four gaps: no images, no self-hosting, no published size and no fine-tuning.

The alternatives range from 287M parameters, for GLiNER2.5-multi-Decide, to 35B-A3B, for Mapika Decider. Jev's size is not disclosed. Encoders such as Von and Laya run on CPU, while 4B adapters fit a 24 GB GPU or a 32 GB Mac. Most open models trail Jev on hard reasoning, according to the report, but several match or beat it on narrow classification. Benchmark comparisons carry caveats: most figures are self-reported; the suites include JevBench, Decision Index, typed-decisions and Respan's behavior benchmark; and scores across suites are not comparable. MarkTechPost noted that the one multi-model third-party table comes from Respan, which sells a competing model.

Cloudflare announced Clef on October 1, 2026. Clef is 27B on Qwen3.8-27B with a vision encoder, and Clef-flash is 9B on Qwen3.5-9B. Both are Apache-2.0 on Hugging Face and served on Workers AI, accepting text, JSON and up to four images per request. Workers AI pricing is $0.24 for Clef and $0.038 for Clef-flash per 1M input tokens. Cloudflare calls the API fully Jev-compatible. In Cloudflare's own run, Clef scored 94.20 BANKING77 F1 versus Jev's 79.74, while Jev won GPQA Diamond, 78.3 versus 48.0.

Liquid AI's d1 family includes hosted d1 and d1-ultra plus two open models. d1-3B has 3.1B parameters on LFM2.5-VL-3B, and d1-omni-600M has 587M parameters and adds audio clips up to 30 seconds. The open weights use the LFM Open License. Liquid's guide calls d1 through TypeSafe's own SDK with a different base URL. Liquid reports d1-3B at 48.57 on Decision Index 0.2.1 and 8 ms latency on an RTX 4090; hosted parameter counts and paid rates are not disclosed.

OpenAI released its Decisions API in public beta on October 6, 2026. It runs only on gpt-6-luna, accepts text plus base64 images and uses predicate, choice and score question types. Pricing is $0.10 per 1M input tokens, with output not billed. The schema is OpenAI's own and not Jev-compatible. OpenAI claims about 10x faster responses than the Responses API and has published no benchmarks.

Perplexity's pplx-decider-v1-27b is a Qwen3.8-27B fine-tune under Apache-2.0. It reads text, JSON and up to eight images per question, and the API allows 128 questions and 255 options per request. Pricing is $0.02 per 1M input tokens for v1.2 at 27B, while a 4B flash-v1.2 costs $0.01. On Perplexity's 11-task panel, it scored 85.71% versus Jev's 84.51%; Jev still won BBH, 94.27 versus 82.80.

meraGPT announced Decider 1 on September 22, 2026. It is hosted, closed-weight and text-only, allowing up to 10 Choice labels and a 4,096-token context, at $0.04 per 1M input tokens. It exposes /v1/systemone and works with the TypeSafe SDK. On meraGPT's typed-decisions run, it scored 0.768 versus Jev's 0.727. An independent review cited an ECE of 0.180, trailing Jev's 0.144.

Respan's Span-01 is a behavior-scoring model rather than a general decision model, returning present, absent or not_observable probabilities. It costs $0.02 per 1M input tokens, and Span-01 Lite is free. Respan claims 0.843 overall F1 on its Behavior Benchmark, but its own blog table shows a lower 0.806, with GPT-6 Sol ahead at 0.885. Size, base model and license are not disclosed.

Standard Thinking's Standard One ships as 3B and 8B models, both open and hosted. The 8B is a LoRA on Ministral-3-8B under Apache-2.0 and reads text and images with a 32,768-token context. It serves /v1/systemone. Hosted text pricing is $0.018 for the 3B and $0.035 for the 8B per 1M tokens. On JevBench public hard, the 8B scored 54.95% versus Jev's 72.07%, while leading on realistic transfer, 90.83% versus 86.67%. Both runs are publisher-measured.