---
title: "AI CTRL Ratecard"
slug: "ai-ctrl-ratecard"
updated: 2026-08-19T10:01:04Z
published: 2026-08-19T10:01:04Z
canonical: "kb.expedient.com/ai-ctrl-ratecard"
excludeFromSearch: true
excludeFromExternalSearch: true
---

> ## Documentation Index
> Fetch the complete documentation index at: https://kb.expedient.com/llms.txt
> Use this file to discover all available pages before exploring further.

# AI CTRL Ratecard

★ Standard Model

Universal Credits powered by GPT-5 mini100Credits=117.65 MTokensBased on standard usage patterns with a 9:1 input-to-output ratio. Credits work across all models — token allocations vary by model capability.OpenAIAnthropicGooglePerplexity⬆ More tokens than GPT-5 miniO

### OpenAI

ChatGPT family of models

| Model | Input Tokens / 100 CR | Output Tokens / 100 CR |
| --- | --- | --- |
| GPT-5 miniStandard General chat, analysis, and light coding | 200M | 25M |
| GPT Image 1.5 Strong image quality, variations, and multimodal workflows at scale | 10M | 1.56M |
| GPT Image 2 Latest image generation with rich detail, edits, and complex scenes | 10M | 1.67M |
| GPT-4.1 Strong coding and long-context instruction following | 25M | 6.25M |
| GPT-4.1 mini Cost-effective coding, extraction, and everyday assistants | 125M | 31.25M |
| GPT-4.1 nano High-volume classification, routing, and low-latency tasks | 500M | 125M |
| GPT-4o Vision and general multimodal assistants | 20M | 5M |
| GPT-4o mini Budget multimodal chat, vision, and simple extraction | 333.33M | 83.33M |
| GPT-5 Reasoning, coding, and general-purpose assistants | 40M | 5M |
| GPT-5 nano Summaries, tagging, and high-volume triage | 1B | 125M |
| GPT-5 Pro Deep analysis and planning with extended thinking | 3.33M | 416.67K |
| GPT-5.1 Everyday coding and agent-style tasks | 40M | 5M |
| GPT-5.2 Coding, tool use, and multimodal workflows | 28.57M | 3.57M |
| GPT-5.2 Pro Very demanding reasoning, long-form analysis, complex decisions | 2.38M | 297.62K |
| GPT-5.3 Codex Agentic coding, refactors, and multi-file software work | 28.57M | 3.57M |
| GPT-5.4 Coding, agents, math, and broad analysis | 20M | 3.33M |
| GPT-5.4 Mini Quick iterations, lighter coding, and tool-heavy workflows | 66.67M | 11.11M |
| GPT-5.4 Nano Snippets, small fixes, classification, and routing | 250M | 40M |
| GPT-5.4 Pro Hardest problems, maximum precision, long-running work | 1.67M | 277.78K |
| GPT-5.5 Flagship for coding, agents, computer-use, and professional work | 10M | 1.67M |
| GPT-5.5 Pro Highest-precision reasoning, audits, and high-stakes professional work | 1.67M | 277.78K |
| GPT-5.6 Luna High-volume extraction, classification, drafts, and cost-sensitive tasks | 250M | 41.67M |
| GPT-5.6 Terra Balanced production coding, analysis, and everyday professional work | 25M | 4.17M |
| GPT-5.6-Sol Frontier flagship (Sol) for hardest reasoning, agents, and professional work | 10M | 1.67M |

A

### Anthropic

Claude Fable, Opus, Sonnet, and Haiku model families

| Model | Input Tokens / 100 CR | Output Tokens / 100 CR |
| --- | --- | --- |
| Claude Fable 5 Frontier Mythos-class reasoning with stronger safeguards for high-risk domains | 5M | 1M |
| Claude Haiku 4.5 High-volume assistants and chat with strong answer quality | 50M | 10M |
| Claude Opus 4.5 Elite reasoning, research depth, and strategic analysis | 10M | 2M |
| Claude Opus 4.6 Hardest agents, complex coding, and very long documents | 10M | 2M |
| Claude Opus 4.7 Stronger coding, vision, and complex multi-step tasks with improved consistency | 10M | 2M |
| Claude Opus 4.8 Most capable Opus for complex reasoning, long-horizon agentic coding, and high-autonomy work | 10M | 2M |
| Claude Opus 5 Deep reasoning, agentic coding, and long-horizon enterprise work | 10M | 2M |
| Claude Sonnet 4.5 Product engineering, coding, and long-context sessions (up to 1M with opt-in context) | 16.67M | 3.33M |
| Claude Sonnet 4.6 Coding, agent planning, and drafting with strong long-context reasoning | 16.67M | 3.33M |
| Claude Sonnet 5 Default Sonnet for agents, coding, and near-Opus quality at lower cost | 25M | 5M |

G

### Google Gemini

Gemini Pro, Flash, and Flash-Lite models

| Model | Input Tokens / 100 CR | Output Tokens / 100 CR |
| --- | --- | --- |
| Gemini 2.5 Flash Chat, summarization, and extraction at large scale | 166.67M | 20M |
| Gemini 2.5 Flash-Lite Routing, translation, and extremely high request volume | 500M | 125M |
| Gemini 2.5 Pro Complex code, deep analysis, and very long context | 40M | 5M |
| Gemini 3 Flash Fast multimodal work, thinking-style tasks, large context | 100M | 16.67M |
| Gemini 3.1 Flash-Lite High-throughput tasks where cost and speed both matter | 200M | 33.33M |
| Gemini 3.1 Pro Agents, coding, and multimodal reasoning (preview) | 25M | 4.17M |
| Gemini 3.5 Flash Fast reasoning, multimodal agents, and large-scale production workloads | 33.33M | 5.56M |
| Gemini 3.7 Flash Workhorse Flash for coding, agents, and production workloads | 66.67M | 13.33M |

P

### Perplexity

Sonar search-grounded models

| Model | Input Tokens / 100 CR | Output Tokens / 100 CR |
| --- | --- | --- |
| Sonar Web-grounded answers | 50M | 50M |
| Sonar Deep Research Long reports, due diligence, and literature-style synthesis | 25M | 6.25M |
| Sonar Pro Multi-step research questions, broad sources, nuanced synthesis | 16.67M | 3.33M |
| Sonar Reasoning Pro Deep multi-step reasoning over retrieved web evidence | 25M | 6.25M |

### Understanding This Rate Card

- Higher-capability models consume more credits per token
- Efficient models stretch your credits further
- All values based on 100 Credits
- **GPT-5 mini** is the baseline standard model
- Green values indicate more tokens than GPT-5 mini
- Rates are subject to change by providers
- Actual usage varies based on conversation patterns

© 2026 Expedient. All rights reserved.  |  AI CTRL Platform Documentation  
Page last updated: **8/19/2026, 6:01:02 AM ET**