Solvex Advisory
AI Benchmarking Portal
Internal Intelligence Tool

AI API Cost Benchmarking
Portal

Real-time cost comparison across OpenAI, Google Gemini, and Anthropic Claude — from flagship to lite models. All costs displayed in South African Rand.

FX Rate
R18.50
per USD
Rate Date
6 June 2026
as at benchmark
Providers
3
OpenAI · Google · Anthropic
Models
12
Flagship to Lite

Cost Comparison

Cost per 1 million tokens in ZAR — sorted, filtered, and saved to your local spec store.

12 models shown
Model ProviderTierInput / 1M tokOutput / 1M tokContext WindowMax OutputSave Specs
Gemini 2.0 Flash LiteUltra-low cost variant
GoogleLite
R0.69$0.0375 USD
R2.77$0.1500 USD
default: 1M
default: 8K
Gemini 2.0 FlashHigh speed, low cost
GoogleLite
R1.39$0.0750 USD
R5.55$0.3000 USD
default: 1M
default: 8K
GPT-4o miniFast, cost-efficient
OpenAILite
R2.77$0.1500 USD
R11.10$0.6000 USD
default: 128K
default: 16K
Claude 3.5 HaikuFast and affordable
AnthropicLite
R14.80$0.8000 USD
R74.00$4.0000 USD
default: 200K
default: 8K
o3-miniEfficient reasoning model
OpenAIStandard
R20.35$1.1000 USD
R81.40$4.4000 USD
default: 200K
default: 100K
Gemini 1.5 ProLong-context standard tier
GoogleStandard
R23.13$1.2500 USD
R92.50$5.0000 USD
default: 2M
default: 8K
GPT-4oMultimodal flagship
OpenAIFlagship
R46.25$2.5000 USD
R185.00$10.0000 USD
default: 128K
default: 16K
Claude 3.7 SonnetExtended thinking support
AnthropicFlagship
R55.50$3.0000 USD
R277.50$15.0000 USD
default: 200K
default: 128K
Claude 3.5 SonnetHigh-intelligence standard
AnthropicFlagship
R55.50$3.0000 USD
R277.50$15.0000 USD
default: 200K
default: 8K
Gemini 2.0 Pro2M token context window
GoogleFlagship
R64.75$3.5000 USD
R194.25$10.5000 USD
default: 2M
default: 8K
o3Advanced reasoning model
OpenAIFlagship
R185.00$10.0000 USD
R740.00$40.0000 USD
default: 200K
default: 100K
Claude 3 OpusMost capable for complex tasks
AnthropicFlagship
R277.50$15.0000 USD
R1387.50$75.0000 USD
default: 200K
default: 4K
Blue = input cost  Dark Blue = output costAll costs: ZAR per 1 million tokens · USD sub-price shown beneathContext Window & Max Output fields are editable — click Save to persist

Cost Calculator

Enter your expected token volumes to estimate total cost per run across all models.

Cheapest Option
Gemini 2.0 Flash Lite
R1.25
Google
Median Cost
GPT-4o
R83.25
OpenAI
Most Expensive
Claude 3 Opus
R555.00
Anthropic
RankModelInput CostOutput CostTotal
#1
GoogleGemini 2.0 Flash Lite
R0.69R0.56R1.25
#2
GoogleGemini 2.0 Flash
R1.39R1.11R2.50
#3
OpenAIGPT-4o mini
R2.77R2.22R5.00
#4
AnthropicClaude 3.5 Haiku
R14.80R14.80R29.60
#5
OpenAIo3-mini
R20.35R16.28R36.63
#6
GoogleGemini 1.5 Pro
R23.13R18.50R41.63
#7
OpenAIGPT-4o
R46.25R37.00R83.25
#8
GoogleGemini 2.0 Pro
R64.75R38.85R103.60
#9
AnthropicClaude 3.7 Sonnet
R55.50R55.50R111.00
#10
AnthropicClaude 3.5 Sonnet
R55.50R55.50R111.00
#11
OpenAIo3
R185.00R148.00R333.00
#12
AnthropicClaude 3 Opus
R277.50R277.50R555.00

Methodology & Assumptions

Reference information for reproducibility and audit purposes.

FX Rate

All USD prices are converted to South African Rand at a fixed rate of R18.50 per USD as at 6 June 2026. This rate is used consistently across all models and calculations.

Pricing Basis

Published API list prices per provider as at 6 June 2026 (on-demand, pay-as-you-go tier). Volume discounts, committed-use contracts, and promotional rates are excluded. Prices are ex-VAT.

Token Definition

Token counts follow each provider's published tokenisation standard. Approximately 750–850 English words equal 1 000 tokens. Input and output are billed separately. Cached input pricing is not reflected here.

Context Window

Maximum combined input + output tokens the model can process in a single API call. Values shown are the maximum published by each provider. Your deployment may impose lower limits based on tier or configuration.

Saved Specifications

Specs are persisted to browser local storage. Values are retained across sessions on the same device and browser. To transfer data, export the contents of the solvex_saved_specs localStorage key.

Internal Use Only

This tool is for internal advisory use by Solvex Advisory. Pricing data should be verified against current provider documentation before finalising client deliverables or procurement decisions. Solvex Advisory accepts no liability for pricing discrepancies.