Model tracker / Decision models / Cloudflare
Clef-flash
Clef-flash is an open-weight decision model from Cloudflare (Apache 2.0, 9B parameters): it answers typed questions, such as yes or no, pick one or pick several, with probabilities instead of writing text. It costs $0.038 per million input tokens (Cloudflare Workers AI, OpenRouter), and output is free. It scores 63.8 (self-hosted; seconds are the board's estimate) on the independent JevBench Capability Score (board v1.7.46), where Jev, the reference, scores 77.1.
- Made by
- Cloudflare
- Weights
- Open, Apache 2.0
- Size
- 9B
- Where
- Cloudflare Workers AI, OpenRouter
- Images
- Yes
- Context
- 24,576 hosted (cut from 64K on 9 Oct); the weights support 256K self-hosted
- Fine-tuning
- Hands-on RL fine-tuning service
Clef-flash price
US dollars per million input tokens. Decision models don't write text, so none of the hosted ones charge for output.
| Where | Input | Output | Notes | Checked |
|---|---|---|---|---|
| Cloudflare Workers AI, OpenRouter | $0.038 | Free | Output is free. Cut from $0.09 on 9 October 2026. | 2026-10-14 |
Clef-flash benchmarks
| Benchmark | Result | What it measures | Reported by |
|---|---|---|---|
| JevBench v1.7.46 | 63.8 | Capability Score: accuracy and calibration on typed decisions, with Jev as the reference system | Independent |
| JevBench time | 0.36 s | Seconds per decision: measured end to end for hosted APIs, the board's estimate for self-hosted models | Independent |
| Reported speed | 39 ms median at launch, before the context cut | Latency on the maker's own hardware | Cloudflare |
Makers' speeds are measured on their own hardware and aren't comparable with each other or with JevBench's timings.
Decision models like Clef-flash
Every other decision model we track, by independent JevBench score.
| Model | Input / M | JevBench | Weights |
|---|---|---|---|
| Clef-flash | $0.038 | 63.8 | Open |
| Jev 1.13 TypeSafe AI | $0.042 | 77.1 | Closed |
| Clef Cloudflare | $0.24 | 75.3 | Open |
| d1 Liquid AI | $0.04 | 74.4 | Closed |
| OpenAI Decisions (GPT-6 Luna) OpenAI | $0.10 | 73.5 | Closed |
| Microsoft-Decision-1 Microsoft | $0.042 | 70.8 | Closed |
| Clef-omni Cloudflare | $0.15 | 58.3 | Open |
| d1-3B Liquid AI | Self-host | 49.5 | Open |
| d1-omni-600M (experimental) Liquid AI | Self-host | 32.8 | Open |
| Laya Convai Innovations | Free | 32.5 | Open |
| CLM-8B Stanford and NVIDIA | Self-host | 20.1 | Open |
| Strands Decider 2B AWS Strands Labs | Self-host | – | Open |
Compare them all in the decision models table →
Clef-flash on dbggr
- Haiku 5.5 and four-cent decision models: the AI price war reaches the bottom tier Saturday edition, 2026-10-10
- Microsoft enters decision models at Jev's price as TypeSafe raises $870M Daily, 2026-10-09
- Claude Haiku 5.5 launches at $0.10 per million tokens, matching GPT-6 Luna Daily, 2026-10-07
- OpenAI publishes 722 AI-written math papers, claiming proofs of the quasi-Riemann hypothesis and Unique Games Daily, 2026-10-06
- Containing agents stopped being the labs' private problem Saturday edition, 2026-10-03
- Stop asking your LLM multiple-choice questions: a guide to decision models Article, 2026-10-02
- Cloudflare and AWS release open-weight decision models to rival Jev Daily, 2026-10-01
Clef-flash FAQ
How much does Clef-flash cost?
Clef-flash costs $0.038 per million input tokens at Cloudflare Workers AI, OpenRouter. Output is free. Cut from $0.09 on 9 October 2026. Checked 14 October 2026.
Is Clef-flash open weight?
Yes. Cloudflare publishes the weights under Apache 2.0, so you can run and fine-tune it yourself.
How fast is Clef-flash?
Cloudflare reports 39 ms median at launch, before the context cut, measured on its own hardware. JevBench gives 0.36 seconds per decision, its estimate for self-hosting.
How does Clef-flash compare with Jev?
On JevBench's Capability Score (board v1.7.46), Clef-flash scores 63.8 against Jev's 77.1. Jev costs $0.042 per million input tokens; Clef-flash costs $0.038. JevBench is a one-person project that uses Jev as its reference, which favours cheap hosted models.