xpu liveBETA
Providers›Novita AI

Novita AI

Visit provider ↗

Novita AI is listed in the xpu live AI provider directory as a inference platform. This profile brings together the models linked to the provider in our catalog, along with documentation and source references. Use it to compare API access and identify the exact model route for your workload.

Inference platform with language models and image, video, text-to-speech and voice-cloning APIs.

Novita AI models and API access

The table below shows a curated selection of linked models rather than a complete list of everything Novita AI offers. Open a model profile to review its recorded capabilities, context limits and provider-specific deployment details. A model name alone does not identify the endpoint version, pricing tier or available features.

Check the Novita AI documentation for authentication, supported request formats and service limits. Regional access and account permissions can differ between routes. Confirm the exact model identifier before integrating an endpoint, especially when a listing is marked preview, legacy or unavailable.

How to compare Novita AI API pricing

Read each rate together with its billing unit and tariff conditions. Text endpoints typically separate input and output token charges, while media routes may bill by image, generated second, audio minute or request. These units measure different work and should not be compared as if they were interchangeable.

For a text workload, estimate input and output volume separately. Include conversation history, retrieved documents and tool responses in your input estimate. Batch processing, caching and priority settings may affect the effective rate. Match these settings when comparing Novita AI with another provider.

A headline price is not a complete project budget. Consider repeated generations, retries, storage and any platform charges relevant to your endpoint. A blank price means the catalog has no verified numeric rate for that offer; it does not mean the service is free.

Source checks and provider selection

Novita AI catalog entries include source references and observation dates where recorded. Availability and prices can change after a check. Review the provider’s current terms before production use and test representative requests for quality, latency and reliability.

Use the official catalog reference to review the provider’s own listing. The model comparison table can include the best collected tariff across providers; open a model’s pricing section to inspect the specific offer from Novita AI.

Browse other AI API providers or return to the AI model rankings and pricing directory. Choose an endpoint based on the required capabilities, expected usage and service conditions, alongside its advertised unit rate.

12 linked modelsAPI documentation ↗Official catalog ↗Checked 2026-09-28

Models on Novita AI 12

A curated selection, with links to other providers for the same model.

Verified catalog sources
Ranked by Arena user preference

Arena Score — Text / overall ↗ · Higher is better · Checked 2026-11-22T14:00:05.877Z

Arena / LMArena data is licensed under CC BY 4.0. Selection and catalog matching by XPU LIVE. Human preferences; not the former Intelligence Index. Automatically checked every six hours. Reference rank and catalog rank are separate; original precision and tied ranks are retained. Unverified and non-language models appear last without a rank. Methodology ↗

Source published 2026-11-19 · CC BY 4.0 license

12 models

RankModelArena ScoreBest priceInput → outputProviders
#29Catalog #14Kimi K3Moonshot AI1474.7Kimi K3 (Max)
$0.192 in / $16 out
USD / 1M tokensOpenRouter

Router catalog · Lowest listed input price; output is from the same tariff.

OpenRouter route catalog price, not a direct-provider price. Routing, modality and endpoint conditions can change the actual bill.

Compare prices →
Router catalogUpdated 2026-11-22 · 15:43 UTC
text, image, video → textDeepInfraFireworks AINovita AIOpenRouterTogether AI
#36Catalog #17GLM 5.2Z Ai1469.5GLM-5.2 (Max)
$0.322 in / $0.989 out
USD / 1M tokensOpenRouter

Batch / Router catalog · Lowest listed input price; output is from the same tariff.

Batch processing tariff; not a synchronous Standard request. OpenRouter route catalog price, not a direct-provider price. Routing, modality and endpoint conditions can change the actual bill.

Compare prices →
Batch / Router catalogUpdated 2026-11-22 · 15:43 UTC
text → textCloudflare Workers AINovita AIOpenRouter
#104Catalog #32MiniMax M3MiniMax1426.5MiniMax-M3
$0.224 in / $0.88 out
USD / 1M tokensOpenRouter

Batch / Router catalog · Lowest listed input price; output is from the same tariff.

Batch processing tariff; not a synchronous Standard request. OpenRouter route catalog price, not a direct-provider price. Routing, modality and endpoint conditions can change the actual bill.

Compare prices →
Batch / Router catalogUpdated 2026-11-22 · 15:43 UTC
text, image, video → textNovita AIOpenRouter
—DeepSeek V4 Pro 0813Deepseek—Not verified
$0.66 in / $1.98 out
USD / 1M tokensDeepSeek

Off-peak · Lowest listed input price; output is from the same tariff.

Time-of-day tariff; consult official source for UTC peak/off-peak windows and alias behavior.

Compare prices →
Off-peakUpdated 2026-11-22 · 15:43 UTC
text → textCloudflare Workers AIDeepSeekNovita AIOpenRouter
—Fish Audio TTSFish Audio—Not comparableSee pricing options →text → audioNovita AI
—Hailuo 2.3MiniMax—Not comparable
From $0.35
USD / generationSegmind

Resolution=768p / Duration (seconds)=6 · Starting tariff; final cost depends on the selected settings.

Total listed price per generation for exactly these parameters.

Compare prices →
Resolution=768p / Duration (seconds)=6Updated 2026-11-22 · 15:43 UTC
text, image → videoNovita AISegmind
—Hailuo 2.3 FastMiniMaxDiscontinued / unsupported · Novita endpoint only—Not comparableNo active offertext, image → videoNovita AI
—Ling 3.1 Flash 1019InclusionAI—Not verified
$0.3 in / $0.9 out
USD / 1M tokensNovita AI

Standard · Lowest listed input price; output is from the same tariff.

Decimal USD per million fields from the provider; tiered entries require separate mapping.

Compare prices →
StandardUpdated 2026-11-22 · 15:43 UTC
text → textNovita AI
—MiniMax Voice CloningMiniMax—Not comparableSee pricing options →audio → voice profileNovita AI
—Qwen-ImageAlibaba / Qwen—Not comparable
From $0.0058
USD / imageTogether AI

Estimated starting image price · Starting tariff; final cost depends on the selected settings.

Provider estimate, not a fixed per-image tariff. The upstream charge varies with resolution, quality and request parameters.

Compare prices →
Estimated starting image priceUpdated 2026-11-22 · 15:43 UTC
text → imageNovita AITogether AI
—Qwen-Image EditAlibaba / Qwen—Not comparable
From $0.0043
USD / gpu secondSegmind

Serverless · Starting tariff; final cost depends on the selected settings.

Billed per GPU compute second, not generated media duration. Runtime varies.

Compare prices →
ServerlessUpdated 2026-11-22 · 15:43 UTC
text, image → imageNovita AISegmind
—Qwen3.8 FlashAlibaba / Qwen—Not verified
$0.15 in / $0.47 out
USD / 1M tokensNovita AI

Standard · Lowest listed input price; output is from the same tariff.

Decimal USD per million fields from the provider; tiered entries require separate mapping.

Compare prices →
StandardUpdated 2026-11-22 · 15:43 UTC
text, image, video → textNovita AIOpenRouter

Catalog entries link to their specification and availability sources. This is a curated selection, not every model each provider offers. Best prices are the lowest collected USD tariffs, not a total-cost estimate. Token offers are selected by input price, with output from the same tariff; workload mix may change the cheapest provider. Listed variants may require batch processing, off-peak hours, or specific media settings. Price coverage varies by provider; collected tariffs include their source and observation time. Benchmark scores are a dated snapshot, not a live feed or a universal measure of quality. They apply to the shown evaluation configuration; listed prices may use different settings. Models without verified scores are not assumed to be weaker.