Cheapest provider for a model

Prices your workload against every endpoint of one model and ranks them. Gateways and aggregators are left out unless you include them. Everything runs in your browser; the link keeps your inputs.

Start with a popular model

GLM-5.3widest coverage in the catalogue — 83 providersZ.ai (Zhipu AI) · 1.05M context · from $0.75/1M in GLM-5.3-Flashthe cheap workhorse, 72 providersZ.ai (Zhipu AI) · 1.05M context · from $0.099/1M in Kimi K2.651 providers, open weightsMoonshot AI · 262K context · from $0.45/1M in Claude Sonnet 5.5Anthropic flagship-tier, 17 providersAnthropic · 1M context · from $2.00/1M in Claude Haiku 5.5small and fast, 16 providersAnthropic · 1M context · from $0.10/1M in GPT-6.1 SolOpenAI, 16 providersOpenAI · 1.05M context · from $2.00/1M in Gemini 3.5 Flash12 providersGoogle · 1.05M context · from $1.50/1M in Qwen3 Coder Nextcoding-tuned, 11 providersAlibaba (Qwen) · 262K context · from $0.12/1M in

Model

Workload (per request)

Input tokensOutput tokens

Cache hit %Reasoning tokensPart of output. Billed separately only where a reasoning price exists.

Service tierEndpoints with no price for this tier fall back to standard, and say so.Billing region (regional uplift)No endpoint in the data publishes a regional uplift.Request time, UTC (for time-of-day prices)Empty: the standard price is used and any discount windows are listed.

Constraints

Regions (any of)

Matches the region the endpoint states, or its provider's datacenters when the endpoint states none. Global matches Global only.

Include endpoints whose region is unknownMinimum precisionInclude endpoints whose quantization is unknown or not rankableMinimum context (tokens)

Required capabilities

Required parameters

Include endpoints where a required capability or parameter is unknownZero data retentionInclude endpoints where ZDR is unknown (only confirmed ones are known)Include deprecated endpointsInclude gateways in the ranking

Loading…