AI Model Market

$0.50~~/1M in~~ · $3.00~~/1M out~~ ↓75% View →

Gemini 3.1 Pro

gemini-3.1-pro

Gemini 3.1 Pro is the next-generation flagship multimodal model from Google, with a very long context window and top-tier reasoning and coding. Fully OpenAI-API compatible: set model to gemini-3.1-pro in your existing SDK. Low-reasoning tier by default (faster and cheaper); pass reasoning_effort=high for the high tier (billed at the Preview rate). Responses carry a reasoning_content field so you can see the thinking (streamed chunk by chunk as well).

$0.60~~/1M in~~ · $3.60~~/1M out~~ ↓7% View →

Gemini 3.8 Flash

gemini-3.8-flash

Gemini 3.8 Flash is the latest Flash generation from Google (released September 2026), built for long-running coding, agent workflows and complex reasoning: 1M-token context, up to 64K output, image input. It uses a dynamic thinking budget -- quick answers for simple questions, more thinking for hard ones. Fully OpenAI-API compatible: set model to gemini-3.8-flash.

Gemini 4 Argon Not yet open

gemini-4-argon Beta

Gemini 4 Argon Beta: three thinking levels and a shared million-token context

— View →

Claude

Anthropic

$2.50~~/1M in~~ · $12.50~~/1M out~~ ↓38% View →

Claude Opus 5.5

claude-opus-5-5

Claude Opus 5.5 is the newest model in Anthropic's Opus line and the successor to Claude Opus 5, built for long-horizon agentic coding and knowledge work: a 1M-token context window, up to 128K output tokens per request, and adaptive thinking that is always on. It supports image input, tool calling, prompt caching and streaming. NezhaGate serves it through both the OpenAI-compatible API and the native Anthropic Messages API (/anthropic), so Claude Code connects directly; the five effort levels (low to max), thinking summaries, structured output and PDF input are available on the native API. Pay as you go; failed calls are never billed.

$1.50~~/1M in~~ · $7.50~~/1M out~~ ↓25% View →

Claude Sonnet 5.5

claude-sonnet-5-5

Claude Sonnet 5.5 is the newest model in Anthropic's Sonnet line and the successor to Claude Sonnet 5, which Anthropic calls its best combination of speed and intelligence: output more than 30% faster than Sonnet 5, fewer tokens and tool calls for the same task, and 70.6% on Terminal-Bench 4.0. It has a 1M-token context window, up to 128K output tokens per request and adaptive thinking, with image input, tool calling, prompt caching and streaming. NezhaGate serves it through both the OpenAI-compatible API and the native Anthropic Messages API (/anthropic), so Claude Code connects directly; effort control, thinking summaries, structured output and PDF input are available on the native API. Pay as you go; failed calls are never billed.

$4.00~~/1M in~~ · $20.00~~/1M out~~ ↓20% View →

Claude Opus 5

claude-opus-5

Claude Opus 5 is the next-generation flagship from Anthropic, with adaptive thinking on by default. It excels at complex reasoning, code engineering and agent orchestration, and supports image understanding, tool use and prompt caching. Fully OpenAI-API compatible: set model to claude-opus-5; or use the native Anthropic Messages API so Claude Code connects directly.

$8.00~~/1M in~~ · $40.00~~/1M out~~ ↓20% View →

Claude Fable 5

claude-fable-5

Claude Fable 5 belongs to the Anthropic Fable series: same generation and capability base as Opus 5 (adaptive thinking, tool use, image understanding, prompt caching), with a finer touch and richer emotional layering in Chinese narrative and long-form writing. Fully OpenAI-API compatible: set model to claude-fable-5; the native Anthropic Messages API is supported as well.

$1.50~~/1M in~~ · $7.50~~/1M out~~ ↓50% View →

Claude Sonnet 4.6

claude-sonnet-4-6

Claude Sonnet 4.6 strikes an excellent balance between capability and speed, excelling at everyday chat, code and agents at a great price. Fully OpenAI-API compatible: set model to claude-sonnet-4-6 and you are connected.

Grok

xAI

$0.30~~/1M in~~ · $0.90~~/1M out~~ ↓85% View →

Grok 4.7

grok-4.7

Grok 4.7 is xAI's flagship reasoning model, strong at complex reasoning, coding and long-document analysis, with image input. It always thinks before it answers; the reasoning comes back in reasoning_content and reasoning tokens bill at the output rate. Tool calling, JSON Schema structured output and streaming are supported. Fully OpenAI-compatible: set model to grok-4.7. Pay-as-you-go, failed calls never billed, and no long-context surcharge.

DeepSeek

deepseek-v4.1-flash · 0731

DeepSeek V4 Flash

The DeepSeek V4 Flash family: V4.1 is DeepSeek's current Flash release, and 0731 is a pinned V4 Flash snapshot for workloads that need a fixed version and reproducible output. Both cost the same and both support thinking mode and tool calling. Use Model Type to pick a version.

$0.07~~/1M in~~ · $0.30~~/1M out~~ ↓50% View →

Qwen

Alibaba

qwen3.8-max · 0902 · flash

Qwen3.8

The Qwen3.8 family: Max is the flagship of Alibaba Cloud's Qwen 3.8 generation, the strongest at reasoning, coding and agents; 0902 is a pinned Max snapshot for reproducible output, and it caches a repeated long prefix automatically; Flash is the light, fast version with a low unit price for batch jobs and high concurrency. All three always think and support tool calling and image input. Use Model Type to pick a version.

$1.40~~/1M in~~ · $4.20~~/1M out~~ ↓30% View →

GLM

Zhipu AI

$1.12~~/1M in~~ · $3.52~~/1M out~~ ↓20% View →

GLM-5.3

glm-5.3

GLM-5.3 is the new flagship of the GLM-5 series from Z.ai (Zhipu), built for coding, agentic workflows and complex reasoning, with dependable Chinese and English writing. It always thinks before it answers, with the reasoning in reasoning_content, and it supports tool calling and streaming. Fully OpenAI-compatible: set model to glm-5.3. Pay-as-you-go, failed calls never billed, and cached input settles at the cache rate.

GPT Image

OpenAI

gpt-image-2.5 · flare · sunburst

GPT Image 2.5

GPT Image 2.5 is the next-generation OpenAI image model in two style branches: Flare comes back cleaner and smoother, Sunburst grainier with heavier texture and stronger highlights. Identical capabilities and price - only the look differs. Use Model Type to pick one.

$0.005~~/image and up~~ View →

$0.005~~/image and up~~ View →

GPT Image 2

gpt-image-2

GPT Image 2 is built for commercial creative work: article covers, product shots, posters, infographics and illustrations. Supports both text-to-image and image-to-image, 1K / 2K / 4K resolutions and multiple aspect ratios, through the OpenAI-compatible images API.

Nano Banana

Google

$0.020~~/image and up~~ View →

Nano Banana Pro

nano-banana-pro

Nano Banana Pro (Gemini 3 Pro Image) is a flagship image model with stronger detail and composition. Supports text-to-image and image-to-image, excels at typography and complex scenes, and suits commercial covers and high-quality illustration, through the OpenAI-compatible images API.

$0.015~~/image and up~~ View →

Nano Banana 2.1

nano-banana-2.1

Nano Banana 2.1 is Google's October 2026 upgrade of Nano Banana 2 (Gemini 3.1 Flash Image). It keeps Flash-level speed while improving image quality, prompt adherence, in-image text and infographic layout, and wide or panoramic ratios no longer show repeated, tiled patches. One model handles text-to-image and editing, and edits can combine several reference images. 1K / 2K / 4K and many aspect ratios through the OpenAI images API: set model to nano-banana-2.1. Billed per image, failed jobs refunded in full.

$0.015~~/image and up~~ View →

Nano Banana 2

nano-banana-2

Nano Banana 2 (Gemini 3.1 Flash Image) excels at covers, posters, illustrations and product shots. Supports text-to-image and image-to-image, 1K / 2K / 4K resolutions and multiple aspect ratios; fast and cost-efficient, through the OpenAI-compatible images API.

Grok Imagine

xAI

grok-imagine-image · quality

Grok Imagine Image

Grok Imagine Image is xAI's image model in two tiers: Standard is fast and low-priced; Quality holds detail, composition and in-image lettering more reliably. Both do text-to-image and image editing at 1K / 2K. Use Model Type to switch.

$0.005~~/image and up~~ View →

Seedance

ByteDance

Seedance 2.5

seedance-2.5 · 30s Beta

Seedance 2.5: per-second and temporary per-clip tiers. Capabilities and prices depend on the selected model.

$0.158/s View →

seedance-2.0 · fast · clip Beta

Seedance 2.0

The ByteDance Seedance 2.0 family: 2.0 standard renders 480P / 720P / 1080P and Fast is the quickest, 5-15 seconds billed per second; the Clip special tier is $1.00 per clip at 5 / 10 / 15 seconds and accepts photos of real people as references.

$0.150/s View →

$6.1–11.1 ~~/1M video tokens~~ View →

Seedance 2.5 · Token

seedance-2.5-token

For creative workflows that value reliable delivery and responsiveness. Original video links and original-file downloads are provided. Content review applies; material preparation and queue times vary by task.

Kling

Kuaishou

Kling 4.0

kling-4.0

Kling 4.0 is the next-generation video model from Kuaishou's Kling. Send no references for text-to-video, or up to 10 reference images, 5 reference videos and 3 reference audio clips, and say in the prompt what each one is for ("the character in image 1, the setting in image 2, the moves in the reference video"); the model brings them into one shot (images are references, not first frames). Any length from 3 to 30 seconds, 720P or 1080P, landscape 16:9 or portrait 9:16, with sound. Asynchronous: submit, then poll the job id for a re-hosted mp4 link; download it promptly. Billed per second: 720P $0.09/s, 1080P $0.125/s; with reference videos the billed seconds are the output seconds plus the reference-video seconds, and reference audio is not billed; failed jobs are refunded in full.

$0.090/s View →

Veo · Omni

Google

Gemini Omni Flash

gemini-omni-flash

Gemini Omni Flash is Google's next-generation video model, released in 2026. One endpoint, parameters pick the mode: text-to-video with no image, one image as the first frame, two as first and last frame, and with image_role=reference one to three images as subject references. 6, 8 or 10 seconds, 720P or 1080P, landscape 16:9 or portrait 9:16, with native audio, and strong at following a storyboard across several shots. Asynchronous: submit, then poll the job id for a re-hosted mp4 link (kept for 15 days). Billed per second at $0.05/s; failed jobs are refunded in full.

$0.050/s View →

Veo 3.1

veo-3.1

Veo 3.1 is Google's AI video model. One endpoint, parameters pick the type: text-to-video with no reference; one image as the first frame, two as first and last frame; with image_role=reference, one to three images as subject or style references. The Fast tier is open: 720p or 1080p, landscape 16:9 or portrait 9:16, 8 seconds per clip with native audio. Asynchronous: submit, then poll the job id for a re-hosted video link (kept for 15 days). Billed per clip (720p $0.20, 1080p $0.22); failed jobs are refunded in full.

$0.200~~/clip+~~ View →

Wan

Alibaba

Wan 3.0

wan3.0-video

The Alibaba Wan 3.0 video family: Standard and Prime have identical capabilities; they differ only in turnaround time and unit price.

$0.065/s View →

MiniMax

MiniMax H3

minimax-h3

MiniMax H3 (Hailuo 03) video model: 1080P HD output, native audio always generated, 5-15 seconds in one-second steps, six aspect ratios. Text-to-video without a reference image, image-to-video with one (used as the first frame). Asynchronous: submit, then poll the job id for a re-hosted mp4 link (kept for 15 days). Billed per second; failed jobs are refunded in full.

$0.036/s View →

Grok Imagine

xAI