Benchmark · 2026

LLM Token Cost by Provider

What does a million tokens actually cost across the major providers? Claude Opus $15/M input, GPT-4 $10/M, Gemini Pro $3.50/M, Claude Sonnet $3/M, GPT-4o-mini $0.15/M. Output is 3–5× input across all providers.

By the sipi.bot engineering team · Published 2026-07-18 · Methodology

$0.15/Mcheapest capable (GPT-4o-mini)
$15/Mmost expensive (Claude Opus)
100×spread between tiers

Token pricing by model

Pricing as of Q2 2026, per million tokens, USD. Prices are list pricing from provider websites and exclude volume discounts and enterprise deals.

ProviderModelInput ($/M)Output ($/M)Output/Input ratioTier
OpenAIGPT-4o-mini$0.15$0.604.0×Budget
AnthropicClaude Haiku$0.25$1.255.0×Budget
GoogleGemini Flash$0.075$0.304.0×Budget
OpenAIGPT-4o$2.50$10.004.0×Mid
AnthropicClaude Sonnet$3.00$15.005.0×Mid
GoogleGemini Pro$3.50$10.503.0×Mid
OpenAIGPT-4$10.00$30.003.0×Premium
AnthropicClaude Opus$15.00$75.005.0×Premium

The three tiers

Budget tier ($0.075–$0.25/M input)

GPT-4o-mini, Claude Haiku, Gemini Flash. Use these for high-volume, low-stakes work: classification, routing, simple extraction, formatting. An agent making 10,000 classification calls per day at $0.15/M and 500 tokens per call spends $0.75/day — negligible.

Mid tier ($2.50–$3.50/M input)

GPT-4o, Claude Sonnet, Gemini Pro. The workhorses for most production agents — capable enough for tool use, multi-step reasoning, and coding. This is where most production agent spend lands. An agent making 1,000 tool-augmented Q&A calls per day at $3/M and 2,500 tokens per call (input + output) spends ~$7.50/day.

Premium tier ($10–$15/M input)

GPT-4, Claude Opus. Reserve for the hardest tasks — complex research, novel reasoning, high-stakes decisions. An agent making 100 research calls per day at $15/M and 8,000 tokens per call spends ~$12/day. Easy to justify for quality; hard to justify for routine work.

Output tokens dominate cost. Output is 3–5× more expensive than input across every provider. An agent that produces a 2,000-token research report pays 5× more for the output than for the prompt that triggered it. If your agent generates long outputs, output pricing matters more than input pricing — and a model with cheaper output (Gemini Pro at 3×) can beat a model with cheaper input (Claude Sonnet at 5×) even when input pricing looks similar.

Provider pricing trends

Provider token pricing has fallen roughly 40–60% over the last 12 months across all tiers. The budget tier has seen the steepest relative cuts (GPT-4o-mini launched at $0.15/M, roughly 1/10th of GPT-4-mini's launch price). The premium tier has been relatively stable — Claude Opus and GPT-4 remain expensive because the underlying capability is still scarce.

The practical implication: agent cost-per-task is falling, but the runaway tail (the top 1% of tasks that cost 50× the median) is not improving — see cost per task. Cheaper models make routine work cheaper; they do not make retry loops less damaging.

Methodology

  1. Provider list pricing as published on OpenAI, Anthropic, and Google AI websites as of 2026-07-18. Volume discounts, enterprise deals, and cached-prompt pricing are excluded.
  2. Output/Input ratio is calculated from list pricing; providers structure their pricing consistently so the ratio is stable.

Pricing changes frequently — verify current pricing on the provider's website before making decisions based on these numbers. Provider pricing changes often; re-check each provider's pricing page before relying on these figures. Released under CC BY 4.0 — cite sipi.bot.

How to use this benchmark

  1. Right-size your model. Most tasks don't need the premium tier. Move routine work to the budget or mid tier.
  2. Watch output pricing. For long-output agents, output pricing can dominate — pick a model with a low output/input ratio.
  3. Compose with a firewall. Token pricing is per-call. A retry loop multiplies that per-call cost 40×. A velocity rule prevents the multiplication.
Cap your token spend with sipi.bot →