Compare representative API prices from mainstream paid model providers including OpenAI, Claude, Gemini, Qwen, DeepSeek, xAI, Mistral, and Runway. This page is a planning index, not a billing guarantee.
Last checked
Providers change model names, regional availability, free quotas, and discount tiers frequently. Follow the source links before committing production spend.
Prices are copied or normalized from official provider pages only.
Token prices use USD per 1M tokens unless the provider uses per-image, per-second, per-credit, or page units.
For tiered pricing, the table keeps the common entry tier visible and explains the higher tiers in notes.
Discount programs such as batch, cache hits, flex, priority, enterprise commitments, and regional processing can materially change final spend.
Modality coverage
| Provider | Text | Image generation | Video generation | Planning note |
|---|---|---|---|---|
| OpenAI | Yes | Yes | Yes | Token-billed GPT models, GPT Image models, and Sora video models. |
| Anthropic Claude | Yes | No generation API listed | No generation API listed | Claude supports multimodal understanding, but this page covers generation prices only. |
| Google Gemini | Yes | Yes | Yes | Gemini, Imagen, and Veo prices are all listed on the Gemini API pricing page. |
| Alibaba Qwen/Wan | Yes | Yes | Yes | Prices vary by deployment region and model family. |
| DeepSeek | Yes | No generation API listed | No generation API listed | Official USD table covers chat and reasoning token prices. |
| xAI Grok | Yes | Yes | Yes | Grok Imagine covers image and video generation/editing. |
| Mistral | Yes | Consumer plan feature; API rate not itemized here | No generation API listed | The public API table primarily lists text, OCR, voice, and related endpoints. |
| Runway | No general LLM chat API | Yes | Yes | Included as a mainstream creative API rather than a general text LLM provider. |
Text APIs
Use input and output token prices to model chat, coding, extraction, search, agents, and reasoning workloads. Cached input, batch, and regional tiers can change total cost more than the headline model price.
| Provider | Model | Representative price | Billing unit | Notes | Source |
|---|---|---|---|---|---|
| OpenAI | GPT-5.5 | $5.00 input, $0.50 cached input, $30.00 output | Per 1M tokens, short context | Long-context requests and priority processing cost more; batch and flex are lower than standard. | OpenAI |
| OpenAI | GPT-5.4 mini | $0.75 input, $0.075 cached input, $4.50 output | Per 1M tokens | Representative lower-cost OpenAI frontier-family option on standard processing. | OpenAI |
| Anthropic | Claude Opus 4.8 | $5.00 input, $0.50 cache hit, $25.00 output | Per 1M tokens | Cache writes are billed separately at $6.25 per 1M tokens for 5 minutes and $10.00 for 1 hour. | Anthropic |
| Anthropic | Claude Sonnet 4.6 | $3.00 input, $0.30 cache hit, $15.00 output | Per 1M tokens | Balanced Claude tier with prompt caching and batch pricing options. | Anthropic |
| Gemini 2.5 Pro | $1.25 input and $10.00 output up to 200k prompt tokens | Per 1M tokens | Prompts above 200k tokens are listed at $2.50 input and $15.00 output. | ||
| Gemini 2.5 Flash | $0.30 input, $2.50 output | Per 1M tokens | Applies to text, image, and video input; audio input is priced separately. | ||
| Alibaba Qwen | qwen3-max, Global | $0.359-$1.004 input, $1.434-$4.014 output | Per 1M tokens, tiered by prompt size | Global deployment uses US Virginia or Germany Frankfurt endpoints and has no free quota. | Alibaba Cloud |
| Alibaba Qwen | qwen-plus, US/EU/HK | $0.40 input, $1.20 non-thinking output, $4.00 thinking output | Per 1M tokens, up to 256k prompt tokens | The 256k-1M prompt tier is higher at $1.20 input, $3.60 non-thinking output, and $12 thinking output. | Alibaba Cloud |
| DeepSeek | deepseek-chat | $0.07 cache-hit input, $0.27 cache-miss input, $1.10 output | Per 1M tokens | 64k context and 8k max output in the official USD pricing detail table. | DeepSeek |
| DeepSeek | deepseek-reasoner | $0.14 cache-hit input, $0.55 cache-miss input, $2.19 output | Per 1M tokens | 64k context, 32k max CoT tokens, and 8k max output in the official USD table. | DeepSeek |
| xAI | grok-4.3 | $1.25 input, $0.20 cached input, $2.50 output | Per 1M tokens | xAI lists a 1M token context window for grok-4.3 on the Chat API pricing table. | xAI |
| Mistral | Mistral Large 3 | $0.50 input, $1.50 output | Per 1M tokens | Mistral also lists Medium 3.5 at $1.50 input and $7.50 output, and Small 4 at $0.10 input and $0.30 output. | Mistral |
Image APIs
Image pricing is less standardized than text pricing. Some providers charge by output image, some by output tokens, and creative API platforms may use credits.
| Provider | Model | Representative price | Billing unit | Notes | Source |
|---|---|---|---|---|---|
| OpenAI | gpt-image-2 | Image: $8.00 input, $2.00 cached input, $30.00 output; text input: $5.00 | Per 1M tokens | OpenAI image generation is token-billed; use the official calculator for per-image estimates by quality and size. | OpenAI |
| Gemini 2.5 Flash Image | $0.039 per image, plus $0.30 input per 1M text/image tokens | Per image and per 1M input tokens | Batch and flex output are $0.0195 per image; priority output is $0.0702 per image. | ||
| Imagen 4 | $0.02 Fast, $0.04 Standard, $0.06 Ultra | Per image | Paid Gemini API tier pricing for Imagen 4 image generation. | ||
| Alibaba Qwen | qwen-image-plus and qwen-image-2.0 | $0.03 per qwen-image-plus image; $0.035 per qwen-image-2.0 image | Per successfully generated image | International deployment bills only successful image outputs; qwen-image-2.0-pro and qwen-image-max are $0.075 per image. | Alibaba Cloud |
| xAI | Grok Imagine Image | $0.02 output per standard image; $0.05-$0.07 output for quality mode | Per image | Input image pricing is listed separately at $0.002 per image for standard and $0.01 per image for quality mode. | xAI |
| Runway | Gen-4 Image and Gen-4 Image Turbo | $0.05 per 720p image, $0.08 per 1080p image, or $0.02 per turbo image | Credits converted at $0.01 per credit | Runway also resells third-party image models such as gpt_image_2 and gemini_2.5_flash through its API credit system. | Runway |
Video APIs
Video models usually bill per generated second, with price differences for resolution, audio, speed tier, model quality, and batch processing.
| Provider | Model | Representative price | Billing unit | Notes | Source |
|---|---|---|---|---|---|
| OpenAI | Sora 2 | $0.10 per second at 720p | Per generated second | Batch pricing is listed at $0.05 per second. | OpenAI |
| OpenAI | Sora 2 Pro | $0.30 per second at 720p, $0.50 at 1024p, $0.70 at 1080p | Per generated second | Batch prices are half of standard prices on the OpenAI table. | OpenAI |
| Veo 3.1 | $0.40 per second standard with audio; $0.10-$0.30 fast; $0.05-$0.08 lite | Per generated second | Google lists separate 720p, 1080p, and 4K prices where available. | ||
| Alibaba Wan | wan2.6-t2v, Global | $0.086012 per second at 720p, $0.143353 per second at 1080p | Per generated second | Global deployment uses US Virginia or Germany Frankfurt endpoints and has no free quota. | Alibaba Cloud |
| Alibaba Wan | wan2.6-i2v-flash, International | $0.05 per second at 720p with audio, $0.075 at 1080p with audio | Per generated second | No-audio international pricing is $0.025 per second at 720p and $0.0375 at 1080p. | Alibaba Cloud |
| xAI | Grok Imagine Video | $0.05 per second at 480p, $0.07 per second at 720p | Per generated second | Input video is listed at $0.01 per second and image input at $0.002 per image. | xAI |
| Runway | Gen-4 Turbo and Gen-4.5 | $0.05 per second for Gen-4 Turbo, $0.12 per second for Gen-4.5 | Credits converted at $0.01 per credit | Runway also lists Seedance, Veo, HappyHorse, Aleph, and Act-Two video models with model-specific credit rates. | Runway |
Source list
These links should be checked again before budgeting or publishing a production price comparison because model catalogs and discount tiers change often.
FAQ
Providers use different billing units. OpenAI image generation is token-priced, while Google Imagen, Qwen-Image, xAI Imagine, and Runway image endpoints publish per-image or credit-based prices.
Alibaba Model Studio, OpenAI data residency, and other regional deployment options can use different endpoints, data residency rules, and compute pools. Always match the price to the region you actually call.
No. It is a quick planning index. Before production launch, verify the provider page, run a small workload, and compare actual billing export data with your estimate.
