Agnes AI Model Catalog

July 30, 2026 ยท View on GitHub

This catalog summarizes the public Agnes AI model families and the recommended API entry points.

Catalog Status

FieldValue
Catalog version2026.07.30
Last updated2026-07-30 00:00 Asia/Singapore
ScopePublic model, endpoint, quota, and troubleshooting reference
Latest rate-limit updateVideo model RPM updated on 2026-06-28
Change noticeRate limits, subscription quotas, model parameters, and availability may change. Treat the values in this catalog as current reference values, not permanent contractual limits.

API Base URLs

Use CaseBase URL
OpenAI-compatible APIshttps://apihub.agnes-ai.com/v1
Image API roothttps://apihub.agnes-ai.com

Authentication:

Authorization: Bearer YOUR_API_KEY

Text and Agent Models

ModelEndpointCapabilitiesSuggested Use Cases
agnes-1.5-flashPOST /v1/chat/completionsFast chat completions, text generation, image URL input, low-latency inferenceRealtime assistants, content generation, summarization, simple multimodal tasks
agnes-2.5-flashPOST /v1/chat/completionsChat, streaming, tool calling, coding, reasoning, multi-turn dialogue, image understanding, agent workflowsCoding agents, complex developer workflows, tool orchestration, multimodal assistants
agnes-2.0-flashPOST /v1/chat/completionsChat, streaming, tool calling, coding, reasoning, image understanding, agent workflowsDeveloper agents, customer support, coding tasks, workflow automation, multimodal assistants

Text Model Notes

ModelCurrent Reference SpecsNotes
agnes-1.5-flash256K context, 64K max output reference limitRecommended for high-throughput chat and low-latency content workflows.
agnes-2.5-flash512K context, 65.5K max output reference limitFully available to users with Agnes API access. OpenAI-compatible with agnes-2.0-flash; use the same base URL, endpoint, message format, streaming, tool calling, and image URL input. Availability, rate limits, and billing depend on account and API key permissions.
agnes-2.0-flash256K context, 64K max output reference limitThe temporary 1M context window was rolled back in June 2026 for stability. Use this model for coding, reasoning, agents, vision input, streaming, and tool calling.

Image Models

ModelEndpointCapabilitiesSuggested Use Cases
agnes-image-2.0-flashPOST /v1/images/generationsText-to-image, image-to-image, URL output, Base64 outputCreative images, product visuals, posters, image transformation
agnes-image-2.1-flashPOST /v1/images/generationsHigh-density image generation, image editing, URL or Data URI input, flexible image sizesDetailed compositions, marketing assets, character visuals, social media content

Video Models

ModelEndpointCapabilitiesSuggested Use Cases
agnes-video-v2.0POST /v1/videosText-to-video, image-to-video, multi-image video, keyframe animation, async generationStorytelling, marketing videos, product demos, social video, app motion assets

Video result query:

GET https://apihub.agnes-ai.com/agnesapi?video_id=<VIDEO_ID>

Current guidance:

  • Use the returned video_id to poll video results.
  • Do not use task_id for current result polling unless a legacy integration specifically documents that workflow.
  • If a video task stays queued for more than a few minutes, verify that the polling request is using video_id.

Legacy task query format:

GET https://apihub.agnes-ai.com/v1/videos/{task_id}

Current Rate Limits

These values are current public reference values as of 2026-06-28. Base Token Plan quota values were published on 2026-06-22; video model RPM values were updated on 2026-06-28. Use the official platform console as the final source of truth for production traffic planning.

Text Model RPM

Model TypeUser TypePublic Request RPMActual Executable RPM
Text modelsFree / default3020
Text modelsEnterprise6040
Text modelsToken Plan1,0001,000

Image Model RPM

Model TypeUser TypeResolutionPublic Request RPMActual Executable RPM
Image modelsFree / default1K3020
Image modelsFree / default2K2010
Image modelsFree / default3K21
Image modelsFree / default4K11
Image modelsEnterprise1K6040
Image modelsEnterprise2K4020
Image modelsEnterprise3K21
Image modelsEnterprise4K21
Image modelsToken Plan1K120100
Image modelsToken Plan2K12080
Image modelsToken Plan3K21
Image modelsToken Plan4K21

Video Model RPM

Model TypeUser TypePublic Request RPMActual Executable RPM
Video modelsFree / default21
Video modelsEnterprise22
Video modelsToken Plan65

RPM Field Definitions

FieldMeaning
Public Request RPMNumber of requests a user is allowed to initiate per minute.
Actual Executable RPMNumber of requests that can actually be executed per minute after service-side scheduling and capacity constraints.

Current Subscription Quotas

These quota values are current public reference values as of 2026-06-28 and may be adjusted in later pricing or capacity updates.

PlanPriceagnes-2.0-flashagnes-image-2.0/2.1-flashagnes-video-v2.0
Starter$41,500 requests per 5 hours; 15,000 requests per week4,000 images per day500 seconds per day
Plus$107,500 requests per 5 hours; 75,000 requests per week4,000 images per day500 seconds per day
Pro$5030,000 requests per 5 hours; 300,000 requests per week4,000 images per day500 seconds per day

Compatibility Notes

Agnes AI is designed for OpenAI-compatible integrations. For agent clients and coding tools, configure:

Base URL: https://apihub.agnes-ai.com/v1
API Key: YOUR_API_KEY
Chat endpoint: /v1/chat/completions

Recommended model selection:

WorkflowRecommended Model
General chat and content generationagnes-1.5-flash
Coding, reasoning, tool calling, and agent workflowsagnes-2.5-flash
Existing integrations needing a previous-generation compatibility fallbackagnes-2.0-flash
Text-to-image and image editingagnes-image-2.1-flash
Fast image generationagnes-image-2.0-flash
Text-to-video and image-to-videoagnes-video-v2.0

Troubleshooting Reference

StatusMeaningWhat to Check
400Invalid requestRequired fields, parameter types, image URL accessibility, response format placement, model-specific parameter support.
401Authentication failedAPI key value, Authorization: Bearer ... format, environment variable loading, account status.
404Endpoint or resource not foundBase URL, endpoint path, model name, and whether a generated resource ID exists.
429Rate limit exceededCurrent user plan, RPM limits, concurrent requests, retry and backoff behavior.
500Server errorRetry with backoff, reduce payload complexity, verify whether the issue reproduces with a minimal request.
502Upstream gateway errorRetry with backoff and check service status if available.
503Service busy or unavailableRetry later, reduce concurrency, and avoid immediate repeated polling.
520Unknown upstream errorRetry with backoff and capture request metadata for support investigation.
ModelDocs
agnes-1.5-flashhttps://agnes-ai.com/doc/agnes-15-flash
agnes-2.0-flashhttps://agnes-ai.com/doc/agnes-20-flash
agnes-2.5-flashhttps://agnes-ai.com/zh-Hans/docs/agnes-25-flash
agnes-image-2.0-flashhttps://agnes-ai.com/doc/agnes-image-20-flash
agnes-image-2.1-flashhttps://agnes-ai.com/doc/agnes-image-21-flash
agnes-video-v2.0https://agnes-ai.com/doc/agnes-video-v20