Run benchmark

Connect a wallet if you want the address on the record. This beta does not send a transaction. The pipeline is simulated: OpenRouter, Jev, a jury, a report card, then an IQ Labs Code-In inscription.

Models

Claude Opus 5.5
Claude Sonnet 5.5
Claude Haiku 5.5
GPT-6 Astra
GPT-6.1 Sol
GPT-6 Luna
Gemini 3.1 Pro
Gemini 3.8 Flash
Grok 4.7
Grok 4.3
Mistral Large 4
Mistral Medium 3.5
DeepSeek V4 Pro
DeepSeek V4.1 Flash
Qwen 3.8 Max
Qwen 3.7 Plus
Kimi K3
GLM-5.3
Llama 4 Maverick
gpt-oss-120b

Questions

Jury size

3512

Jev by TypeSafeA cheap first pass that returns refusal, hedging, and per-axis scores with confidence, and no text. A smaller LLM jury then writes rationales for a 20% sample and for low-confidence items.

Execution

InstantBatch (about 50% off, queued)

Estimate · simulated

…

This run after cache. Full selection if nothing were cached: ….

Cached answers — · new —

Cached jury — · new —

Inscription about 0.005 SOL, simulated. On-chain cost is separate from the model bill.

Reference at 400 × 20 × 5: naive $155.00, optimized $17.63, with Jev $13.42.

Scaled to this selection: naive —, optimized —, with Jev —.

Wallet

Optional. A live inscription would use the server key, not this wallet.

Bias Detector

A simulated beta beside the Swarm Explorer. Scores, juries, and inscriptions on this site are generated locally and labeled as simulated.

In this product

Overview
Questions
Leaderboard
Models
Inscriptions

Verafy

Swarm Explorer
About Verafy
Help