Cheapest for a need
The cheapest model for what you need
Pick what the job needs and see the models that have it, ranked by what a typical message costs at your billed price. A typical message is about 500 tokens sent and 500 written.
Every price here is the provider's published price plus 10%, the same for every account. We decide each need from the provider's catalog or from a capability probe that saw it work. There is no “best for coding”, because nothing here scores how well a model does a task.
- Reads images204 modelsIt takes images as input, as listed in the provider's catalog.GLM 5.3 Flash by Z.ailess than 1/100 of a cent, $0.04389 in, $0.08624 out per 1M
- 200k tokens and up223 modelsIts context window is at least 200,000 tokens, as listed in the provider's catalog.Ling 3.0 Tiny by inclusionAIless than 1/100 of a cent, $0.0165 in, $0.0495 out per 1M
- 1M tokens and up121 modelsIts context window is at least 1,000,000 tokens, as listed in the provider's catalog.GLM 5.3 Flash by Z.ailess than 1/100 of a cent, $0.04389 in, $0.08624 out per 1M
- Tool calling248 modelsA capability probe here saw its tool calls work, or the provider's catalog declares the tools parameter and no probe has seen it fail.Mistral Nemo by Mistralless than 1/100 of a cent, $0.0319 in, $0.033 out per 1M
- JSON output269 modelsA capability probe here saw its JSON output work, or the provider's catalog declares response_format or structured outputs and no probe has seen it fail.Mistral Nemo by Mistralless than 1/100 of a cent, $0.0319 in, $0.033 out per 1M
- Free16 modelsThe provider publishes a price of zero for both what you send and what it writes. No margin is added to zero.Lyria 3 Clip Preview by Googlefree
Making images? See the cheapest image models by the picture.
What each image model makes, with what one picture costs.
Need something else? The price list filters every model by lab, context, vision, tools and JSON.