# Jev Alternatives: Clef, Perplexity and Strands Decider

> Cloudflare, Perplexity and AWS all shipped Jev rivals on Oct 1, 2026. Prices, sizes, licenses and Cloudflare's benchmarks, side by side.

- Author: Nitish Garg, Founder & CEO, CellCog
- Published: 2026-10-02
- Canonical (HTML): https://cellcog.ai/blog/jev-alternatives/
- Section: Guides / Choosing a platform
- Publisher: CellCog (https://cellcog.ai), the AI employee platform. Blog index for agents: https://cellcog.ai/blog/llms.txt

## Key points

- On October 1, 2026, Cloudflare (Clef and Clef-flash), Perplexity (Decisions API) and AWS Strands Labs (Strands Decider 2B) all released decision models aimed at TypeSafe's Jev.
- Hosted prices per million input tokens: Perplexity $0.04, Jev $0.042, Clef-flash $0.09, Clef $0.24. Strands Decider 2B is self-hosted.
- Clef, Clef-flash and Strands Decider are Apache 2.0; Perplexity's model is on Hugging Face with an Apache 2.0 tag; Jev stays closed and text only.
- On Cloudflare's own runs, Clef-flash answers in a median 38.8 ms against 524.1 ms for Jev, and the Clef models lead most rows, though Jev still wins When2Call and BRIGHT.
- Strands Decider 2B scores 167 of 231 on the public JevBench set and answers in a median 115 ms on an RTX 3090, with training data published.
- No independent benchmark covers all five yet; Perplexity's 85.71% figure appears only in its X post.

## At a glance

- **What are the Jev alternatives?** Cloudflare's Clef (27B) and Clef-flash (9B), Perplexity's pplx-decider-v1-27b behind its Decisions API, and AWS's Strands Decider 2B, all released October 1, 2026.
- **Which is cheapest?** Hosted, Perplexity at $0.04 per million input tokens, just under Jev's $0.042. Strands Decider 2B is free to run yourself.
- **Which is fastest?** On Cloudflare's runs, Clef-flash at a median 38.8 ms, against 524.1 ms for Jev. Those numbers come from Cloudflare, not an independent test.

**Two weeks after TypeSafe's Jev made decision models a category, three big platforms shipped their own on the same day.** On October 1, 2026, Cloudflare released Clef and Clef-flash, Perplexity opened a Decisions API, and AWS's Strands Labs released Strands Decider 2B. All three answer typed questions with probabilities instead of writing text, and all three name Jev as the model they are chasing.

This page is read from each vendor's own docs and launch posts: [Cloudflare's blog](https://blog.cloudflare.com/clef-decision-models/) and [model docs](https://developers.cloudflare.com/workers-ai/models/clef/), [Perplexity's Decisions quickstart](https://docs.perplexity.ai/docs/decisions/quickstart), [the Strands Decider post](https://strandsagents.com/blog/introducing-strands-decider/) and [repository](https://github.com/strands-labs/strands-decider), and [TypeSafe's model docs](https://docs.typesafe.ai/models). For what a decision model is and why Jev spread, see our [Jev record](https://cellcog.ai/blog/jev-typesafe-decision-model/).

## What a decision model does

A decision model takes a state (a ticket, a document, an agent's trace) and a set of questions declared in advance as choices, scores or yes-or-no, and returns a probability for every allowed answer. AWS's post puts the trade plainly: decision models "are faster and more capable at a given size, always produce an answer from the selected options, and can run with very low latency." The cost is just as plain: one parallel pass "makes it significantly worse at solving complex problems than reasoning models", and none of them write text.

## The five side by side

*Table: Decision models as of October 1, 2026, from each vendor's docs*

| Model | Maker | Size | Weights | Input | Hosted price per million input tokens |
|---|---|---|---|---|---|
| Jev 1.13 | TypeSafe AI | Not disclosed | Closed | Text only | $0.042, output free |
| pplx-decider-v1-27b | Perplexity | 27B | On Hugging Face, Apache 2.0 tag | Text, JSON, images | $0.04, output free |
| Clef | Cloudflare | 27B | Apache 2.0 | Text, JSON, images, video | $0.24 |
| Clef-flash | Cloudflare | 9B | Apache 2.0 | Text, JSON, images, video | $0.09 |
| Strands Decider 2B | AWS Strands Labs | 2B | Apache 2.0, with training data and scripts | Text | Self-hosted |

All four new models sit on Qwen bases, per their model cards and posts: Clef and Perplexity's model on Qwen3.8-27B, Clef-flash on Qwen3.5-9B, Strands Decider on Qwen3.5-2B. Context windows differ too: Jev takes 64k tokens per request, Clef 65,536, and Perplexity's request limit is under 262,144.

## Cloudflare's benchmarks against Jev

Cloudflare published the only head-to-head table so far, run on its own evaluation set. It calls the Clef models "smarter, faster, and fully Jev-API compatible", which means existing Jev code can point at them.

*Table: Selected Cloudflare benchmarks, Clef family vs Jev (Cloudflare's runs)*

| Benchmark | Clef | Clef-flash | Jev |
|---|---|---|---|
| BFCL, case exact | 98.47 | 98.76 | 95.75 |
| API-Bank, accuracy | 91.93 | 93.11 | 88.19 |
| BANKING77, macro-F1 | 94.20 | 90.93 | 79.74 |
| CLINC150+OOS, macro-F1 | 97.43 | 66.77 | 89.27 |
| When2Call, accuracy | 72.37 | 65.58 | 80.97 |
| BRIGHT, nDCG@10 | 45.91 | 39.26 | 47.52 |
| Median latency, ms | 209.3 | 38.8 | 524.1 |

Jev still wins two rows in Cloudflare's own table, When2Call and BRIGHT, and Clef-flash drops sharply on CLINC150. The latency gap is partly hosting: Clef runs on Cloudflare's own network.

## Perplexity and AWS

Perplexity's pitch is price and images. At $0.04 per million input tokens it undercuts Jev slightly, and its state can carry images; the docs count about 1,000 input tokens per megapixel. Perplexity's X post cites an 85.71% score across benchmarks; its docs publish no benchmark table, so that number is unverified here.

AWS went the other way: small, local and fully open. Strands Decider 2B answers in a median 115 ms on an RTX 3090, per its README, and scores 167 of 231 (0.723) on the public JevBench set. AWS published the training data and every iteration it tried, which makes it the one to read if you want to build your own.

## What is verified and what is not

- **Prices, sizes and licenses** are read from each vendor's docs and model cards on October 1, 2026.
- **The benchmark table is Cloudflare's.** It chose the tasks and ran every model; no independent comparison of all five exists yet.
- **Perplexity's 85.71%** appears only in its X post.
- **Jev's architecture** is still undisclosed; the three new models published theirs.

## Why it matters

Jev proved the demand; October 1 removed its moat. A buyer now has a cheaper hosted option, two open models with video input, and a 2B model that runs on a laptop, all speaking roughly the same API. For anyone running agents, the decision layer (route this, retry that, stop here) just became a commodity. The value moves to the work the agent does after it decides. For how agent harnesses compare, see our [harness ranking](https://cellcog.ai/blog/best-ai-agent-harnesses/).

## What we are watching for

- An independent benchmark across all five on the same tasks.
- TypeSafe's response: a price cut, open weights, or image input.
- Whether Vercel's AI Gateway adds the new models next to Jev.

## Update log

- October 1, 2026: page opened, from the three launches and TypeSafe's docs.

## Sources

- Cloudflare, [Introducing Clef](https://blog.cloudflare.com/clef-decision-models/), October 1, 2026; [Clef](https://developers.cloudflare.com/workers-ai/models/clef/) and [Clef-flash](https://developers.cloudflare.com/workers-ai/models/clef-flash/) docs.
- Perplexity, [Decisions API quickstart](https://docs.perplexity.ai/docs/decisions/quickstart); [pplx-decider-v1-27b on Hugging Face](https://huggingface.co/perplexity-ai/pplx-decider-v1-27b).
- Strands Agents, [Introducing Strands Decider 2B](https://strandsagents.com/blog/introducing-strands-decider/), October 1, 2026; [GitHub repository](https://github.com/strands-labs/strands-decider).
- TypeSafe AI, [model docs](https://docs.typesafe.ai/models).

## FAQ

**Is Clef a drop-in replacement for Jev?**

Cloudflare says the Clef models are fully Jev-API compatible, so code written for Jev can point at them. Results differ by task: Jev still won two of the benchmarks Cloudflare published.

**Which decision models take images?**

Perplexity's model takes text, JSON and images. Clef and Clef-flash take text, JSON, images and video. Jev and Strands Decider take text only.

**Can I run these models locally?**

Yes for Clef, Clef-flash and Strands Decider, all Apache 2.0 on Hugging Face; Perplexity also posted weights. Strands Decider 2B is built for a laptop or a single GPU.

**Do decision models replace language models?**

No. They cannot write text and are worse at complex problems. They sit around a language model, making the routing, gating and scoring calls an agent makes many times per task.

**Where does CellCog fit?**

We build AI employees for any role. We do not route to any decision model today; we track this layer because it decides what an agent does next, before the expensive model speaks.

## Related

- [Jev, TypeSafe's Decision Model: What It Is, Why It Spread](https://cellcog.ai/blog/jev-typesafe-decision-model/index.md)
- [Best AI Agent Harnesses: October 2026 Rankings](https://cellcog.ai/blog/best-ai-agent-harnesses/index.md)
- [OpenAI DevDay 2026: Everything OpenAI Announced](https://cellcog.ai/blog/openai-devday-2026/index.md)
- [What Is an AI Employee? The 5-Part Test for a Standing AI Worker](https://cellcog.ai/blog/what-is-an-ai-employee/index.md)

## The AI employee for this read

[AI Software Engineer](https://cellcog.ai/ai-employees/ai-software-engineer): I built this page. For what it covers, hire an engineer: it works in your repo behind an approval gate, so nothing reaches your world unclassified.

---

Markdown alternate of https://cellcog.ai/blog/jev-alternatives/. Try CellCog free, no credit card needed: https://cellcog.ai/signup