Text APIs

5-minute cache writes are $12.50/MTok and 1-hour cache writes are $20/MTok.

Fast mode and data residency are priced separately.

Opus 4.7 and later use a newer tokenizer; token counts can differ from older models.

Fast mode has separate premium pricing; US-only inference geography can add a 1.1x multiplier.

OpenAI lists separate Standard, Batch, Flex, and Priority tiers; this row uses Standard short-context pricing.

Regional data residency endpoints may add a 10% uplift for eligible newer models.

Prompts above 200K tokens have higher input, output, and cache prices.

Search and Maps grounding have separate charges after free quotas.

Preview models can change before becoming stable.

Cached input storage is listed as limited-time free on the source page.

Cached input storage is listed as limited-time free.

Batch API pricing is 50% lower for input and output tokens.

Kimi pricing page lists cache-hit, input, and output prices.

Official page marks this as permanent 50% off compared with crossed-out list price.

Input tokens above 512K may have availability limits.

Supports text, image, video input, thinking and non-thinking modes.

Long-context standard pricing is higher; Batch is discounted where available.

Priority pricing is listed separately on OpenAI pricing.

US-only inference geography adds a 1.1x multiplier for Sonnet 4.6 and later models.

Cached input storage is listed as limited-time free.

Good row for budget-sensitive routing comparisons.

Displayed as Free in the official pricing table; availability can change.

Prices are per 1M tokens.

Batch pricing is 50% lower in the official table.

Good reference point for lightweight classification, extraction, and chat routing.

No cached-input price is listed for this pro row in the flagship pricing table.

Prompt cache read and write are separate pricing fields.

High-speed tier doubles input and output price compared with standard.

Longer prompt tiers are priced higher in the official table.

Prompt cache writes are listed separately from cache hits.

Legacy model retained for users comparing older production integrations.

High-speed tier doubles input and output price compared with standard.

Server-side tools are charged separately from token usage.

Global endpoint and storage are in US Virginia or Germany Frankfurt.

Cached input storage is listed as limited-time free.

Useful Claude baseline for high-volume support and extraction workloads.

Batch processing gets a 50% discount.

Audio input is priced separately on Google pricing.

Mistral pricing is per million input and output tokens.

Prompts above 200K tokens have higher prices.

Useful when comparing ChatGPT chat-latest against flagship model APIs.

Official table marks Mythos 5 as limited availability.

Grounding and Maps pricing are separate from token billing.

Currently described by xAI as early access.

Listed in xAI Chat API pricing with the same token rates as grok-4.3.

Reasoning tokens are billed under the model token rates.

Batch pricing can vary by model detail page.

Open model listed on Mistral pricing.

Use for reasoning comparisons against general-purpose models.

Batch processing gets a 50% discount.

Best for low-cost routing and lightweight agent steps.

Low-cost open model in Mistral pricing.

Cohere lists Aya Expanse API pricing in the official pricing FAQ.

Cohere marks these as legacy model prices for existing customers.

Listed in Cohere pricing FAQ as legacy pricing.

Listed in Cohere pricing FAQ as legacy pricing.

Listed in Cohere pricing FAQ as legacy pricing.

Listed in Cohere pricing FAQ as legacy pricing.

Automatic context caching uses cache-hit and cache-miss input prices.

Reasoning output and CoT behavior can change real workload cost.

Higher prompt-size tiers increase input and output prices.

Higher prompt tiers are listed separately up to 256K tokens.

International inference is dynamically scheduled globally excluding Chinese Mainland.

Listed under QwQ open source model pricing.

Chinese Mainland deployment only according to official table.

Useful for math-specific API cost comparisons.

Limited-time promotion is mentioned on the official page.

Cached input storage is listed as limited-time free.

Prices are per 1M tokens.

Cached input storage is listed as limited-time free.

Prices are per 1M tokens.

No cached-input price is listed for this row.

Displayed as Free in the official pricing table; availability can change.