Token Cost Index by Hardcut
Updated May 2026

LLM prices have fallen 97% since GPT-4 launched.

Here's where the market stands today.

Cost-per-Task Calculator
Pick a workload and see what 1 request costs at each provider's current flagship · plus monthly cost at typical volumes
Provider · Current Model Per Request per 1K req/mo per 10K req/mo per 100K req/mo vs cheapest
Price Trajectory by Provider
Output-token $/1M for the best-value flagship at each provider
Range
Metric
Scale

Provider Snapshot

Current flagship, trajectory, and key stats per provider
Price Floor Over Time
Cheapest available model on the market each month
Current Price Distribution
Latest model price for each provider, sorted
Context Window Evolution
Max input window per flagship over time — log scale, in tokens
Value Frontier — Intelligence vs Cost
AA Intelligence Index vs output price · bottom-right = smartest for cheapest · Source: Artificial Analysis
Industry Timeline
— · chart annotations show high-impact events; full set below
Filter
Provider Volatility & Trajectory
Pricing stability, swing range, and net change over each provider's lifetime
Provider First Price Current All-Time Low All-Time High Spread (×) Net Change # Releases
Full Cost Breakdown
Every flagship-model price point in the dataset
Date↕ Provider↕ Model↕ $/1M In↕ $/1M Out↕ Ctx↕ AA Intel.↕ TKI↕ Δ Out↕ Held↕
TKI baseline: GPT-4 at launch (March 2023) = 100 • Output-token cost for best-value flagship per provider
Sources: PricePerToken · BenchLM · TokenMix · Intelligence scores: Artificial Analysis  |  Built by Hardcut  |  Press T to cycle theme