Run real benchmarks. See real scores.
Pay only for what completes.

TAB is usage-priced with prepaid credits. No monthly plans required. Every run shows an estimate and a hard spending cap before you start.

Add Credits →

How It Works

1

Add credits ($10 minimum)

Buy once, use anytime. Credits never expire.

2

Run benchmarks

See the cost estimate before you start. Set a Max Spend cap.

3

Pay only for completed cases

Failed cases are automatically refunded to your balance.

Rate Card

Per-case pricing. Multiply by the model tier for your total.

Benchmark Type Per Case Runtime
Text (prompt / response) $0.03 × model tier
Tool-Use (API interaction) $0.10 × model tier
Browser (navigation tasks) $0.25 × model tier + $0.02/min
Sandbox (code execution) $0.40 × model tier + $0.03/min
Verification (API lookup) $0.01 flat
Security Screening FREE Always free

Model Tier Multipliers

Your per-case rate is multiplied by the tier of the model you choose.

Tier Multiplier Example Models
Core Gemini Flash, GPT-4o-mini, Llama 4, DeepSeek V3
Pro GPT-4o, Claude Sonnet, Gemini Pro
Premium Claude Opus 4.5, GPT-5, Gemini Pro
Ultra 10× Claude Opus 4.6, GPT-5.2 Pro, o3-pro

Example: 50 tool-use cases on a Pro model = 50 × $0.10 × 2 = $10.00

Sell your verified agents. Keep more when you verify more.

Our verification rebate model rewards developers who invest in quality.

Verification Level Platform Fee You Keep
Unverified 25% 75%
Security Screened 22% 78%
Core Benchmarked 18% 82%
Fully Certified 15% 85%

Bring Your Own Keys

Bring your own model API keys. TAB charges for evaluation and infrastructure only — your provider bills model inference directly to you.

No Surprises Promise

📊

Every run shows an estimate before you start.

🛑

Your Max Spend cap is a hard stop — TAB will never charge beyond it.

You only pay for completed benchmark cases.

🔄

Failed cases are automatically credited back to your balance.

Frequently Asked Questions

Do credits expire?

Purchased credits never expire. Promotional credits have a 90-day window.

What's the minimum top-up?

$10.00 — enough to run hundreds of text benchmark cases.

What if a benchmark run fails?

Failed cases are automatically refunded to your balance. You only pay for completed work.

Can I use my own API keys?

Yes. BYOK users pay only for TAB's evaluation and infrastructure — your provider bills model inference separately.

When will prices change?

TAB launched with a 60-day price freeze. Prices are locked until May 4, 2026.

Ready to start?

Add Credits → Browse Verified Agents →