AI cost desk
AI model pricing
Compare API prices, context windows, and multimodal support before choosing a model.
Prices are listed per 1M tokens unless a provider states otherwise. Check the source before making purchase decisions.
8
8 providers
8
Every public row links back to a provider page.
2026-07-04
Corrections go through admin review before public data changes.
Use these as starting points, then open the provider source before committing spend.
$0.435 input / 1M
deepseek-v4-pro
DeepSeek is the cheapest USD-denominated input row in this seed set.
1M context
gemini-3.1-pro-preview
Google Gemini has the widest context window in the current seed set.
$0.95 input / 1M
kimi-k2.6
Kimi is the lowest-input-price video-capable row in this seed set.
$0.003625 cached input / 1M
deepseek-v4-pro
DeepSeek has the lowest cached-input row where a provider publishes it.
Seed pricing table
1 filtered rows. Export includes source URLs and localized notes.
| Model | Provider | Input / 1M | Output / 1M | Cached input | Context | Support | Updated | Source |
|---|---|---|---|---|---|---|---|---|
kimi-k2.6 Official model page lists 256K context and multimodal support; pricing seed follows Kimi's public K2.6 API rate card. | Kimi | $0.95 | $4 | $0.16 | 262K | APIImage inputVideo input | 2026-07-04 | Source |
Use these field notes with the table when you are comparing launch access, agent cost, or image-model tradeoffs.
Claude Sonnet 5: agent cost before migration
A practical note on token budgets, migration checks, and why agent workflows need cost review before switching models.
Read noteGPT-5.6 limited preview: access signals
A launch-readiness read on why access limits, partner rollout, and API availability matter beyond headline model names.
Read noteNano Banana 2 Lite: image workflow tradeoffs
A creator workflow note for judging speed, image quality, and multimodal use before adding a visual model to the stack.
Read noteSubmit a price correction
Found a changed price or a better source? Send it for admin review.