The right model for
Text
Arena score vs leader98.1%1475 Arena score
Token cost savings
99.4%input99.6%output
Choose GLM-5.3 Flash for high-volume text: 98.1% of the leader’s Arena score with over 99% lower input and output costs.
AI infrastructure, with perspective
The right AI model
for every workload.
Access leading AI models through one API.
Maximum performance or better economics.
Make the choice that fits your workload.
Choose your workload · Your API key · Run your model
ONE INTEGRATION.
LEADING MODEL FAMILIES.
AnthropicZ.aiMoonshot AIOpenAIQwenWanMiniMax
Start with your workload
What are you building?
Eight capabilities. A considered shortlist.
Start with the task, then choose the model.
Text
Models for conversational applications, reasoning, writing, translation and summarization.
Customer support · AI assistants · Translation · Summarization
Coding
Models for code generation, debugging, review and software development workflows.
Code generation · Code review · Refactoring · Developer assistants
Vision
Models that turn images and visual context into useful text and structured understanding.
Image understanding · Visual search · Quality review · Chart interpretation
Documents
Models for extracting, summarizing and reasoning over document content.
Document extraction · Report summaries · Contract review · Knowledge workflows
Image Generation
Models for creating images from a text description and creative direction.
Product visuals · Campaign assets · Concept art · Creative variants
Image Editing
Models for revising existing images with clear creative instructions.
Product edits · Background changes · Visual revisions · Asset adaptation
Text → Video
Models for turning written concepts into short video assets.
Motion concepts · Campaign clips · Storyboards · Creative testing
Image → Video
Models for adding motion to existing images and visual concepts.
Product animation · Motion variants · Visual storytelling · Social clips
A clear choice, not a crowded catalog
Performance meets value.
The strongest model when quality leads.
The smartest tradeoff when economics matter.
Best Performance
Maximum benchmark performance.
A
Claude Fable 5 High
AnthropicArena score vs leader100%1504 Arena scoreThe leader is the 100% reference
API price
$10.00Input / 1M tokens
$50.00Output / 1M tokens
Choose this when the highest Arena score comes first.
Leads the supplied Arena text comparison. Choose it for reasoning, writing and assistants when benchmark performance is the priority.
Best Value
A smarter performance-to-cost balance.
Z
GLM-5.3 Flash
Z.aiArena score vs leader98.1%1475 Arena scoreShare of the leader’s Arena score
Token cost savings
−99.4%Saves on input cost
$0.06Input / 1M tokens
−99.6%Saves on output cost
$0.20Output / 1M tokens
98.1% of the leader’s Arena score · 99.4% lower input cost · 99.6% lower output cost
A small supplied Arena score gap with lower input and output token prices. Consider it for assistants, support and summarization at volume.
Arena · supplied snapshot·Supplied Arena snapshot · date pending·How we compare
Input and output savings use the separate supplied token rates. Combined cost uses 1M combined tokens: 75% input / 25% output; your workload mix changes the combined savings. Arena scores and prices are from the supplied snapshot. The leaderboard source and snapshot date still need confirmation.
See Text ModelsBest Performance
Maximum benchmark performance.
A
Claude Fable 5 High
AnthropicArena score vs leader100%1551 Arena scoreThe leader is the 100% reference
API price
$10.00Input / 1M tokens
$50.00Output / 1M tokens
Choose this when the highest Arena score comes first.
Leads the supplied Arena coding comparison. Prioritize it for difficult generation, review and debugging workloads.
Best Value
A smarter performance-to-cost balance.
Z
GLM-5.3 Flash
Z.aiArena score vs leader98.3%1525 Arena scoreShare of the leader’s Arena score
Token cost savings
−99.4%Saves on input cost
$0.06Input / 1M tokens
−99.6%Saves on output cost
$0.20Output / 1M tokens
98.3% of the leader’s Arena score · 99.4% lower input cost · 99.6% lower output cost
A small supplied Arena coding score gap with lower input and output token prices for repeated developer workflows.
Arena · supplied snapshot·Supplied Arena snapshot · date pending·How we compare
Input and output savings use the separate supplied token rates. Combined cost uses 1M combined tokens: 75% input / 25% output; your workload mix changes the combined savings. Arena scores and prices are from the supplied snapshot. The leaderboard source and snapshot date still need confirmation.
See Coding ModelsBest Performance
Maximum benchmark performance.
A
Claude Fable 5 High
AnthropicArena score vs leader100%1496 Arena scoreThe leader is the 100% reference
API price
$10.00Input / 1M tokens
$50.00Output / 1M tokens
Choose this when the highest Arena score comes first.
Leads the supplied Arena document comparison. Prioritize it when document benchmark performance matters most.
Best Value
A smarter performance-to-cost balance.
K
Kimi K2.6
Moonshot AIArena score vs leader97%1451 Arena scoreShare of the leader’s Arena score
Token cost savings
−90.5%Saves on input cost
$0.95Input / 1M tokens
−92%Saves on output cost
$4.00Output / 1M tokens
97% of the leader’s Arena score · 90.5% lower input cost · 92% lower output cost
Kimi is the current supplied value recommendation for documents, with lower input and output token reference prices.
Arena · supplied snapshot·Supplied Arena snapshot · date pending·How we compare
Input and output savings use the separate supplied token rates. Combined cost uses 1M combined tokens: 75% input / 25% output; your workload mix changes the combined savings. Arena scores and prices are from the supplied snapshot. The leaderboard source and snapshot date still need confirmation.
See Documents ModelsStart with your workload
Choose your model.
Make it work.
Choose a workload, get your API key,
then replace the key in the model test and run.
Choose your workload.
Compare models for the task you want to build.
Get your API key.
Get API access to receive your own Roten key.
Replace the API key and run.
Use your own API key in the model test. Everything else is already configured.
Choose intelligently
Performance and cost, compared together. Make the tradeoff with the evidence in front of you.
One API
Access supported model families with your Roten API key.
Production pricing
Usage-based economics for companies with meaningful AI consumption.
A local relationship
Contracts, invoicing and commercial support for your business.
Built for production
Running AI at scale?
Talk to us about volume pricing, model selection and migration.