LLMFly AI

PRICING EXPLAINED

LLM API Pricing 2026: GPT, Claude, Gemini & Grok

Compare GPT, Claude, Gemini and Grok API prices per million tokens. Check LLMFly AI input, output and cache rates alongside provider reference prices, then estimate your API costs with the model pricing calculators.

Budget formulaInput + output + retriesConfirm against a real usage record
01

Model rates

Filter 32 models across 4 groups. All amounts are USD per 1M tokens.

Open live Model Plaza ↗
Model group

Showing 7 models

openai

ChatGPT(Codex)

70% off · billing multiplier 0.3x

7 / 7 models
Model / tierLLMFly AIOfficial priceSavings vs. official / 1M
GPT-6 AstraAPI ID: gpt-6-astraLLMFly: ≤272K / >272K · Official: ≤272K / >272K · whole request
≤272K
Input$3Output$15Cache write$3.75Cache read$0.3
>272K
Input$6Output$22.5Cache write$7.5Cache read$0.6
≤272K
Input$10Output$50Cache write$12.5Cache read$1
>272K
Input$20Output$75Cache write$25Cache read$2
70% OFF
≤272KIn save $7Out save $35Write save $8.75Read save $0.7
>272KIn save $14Out save $52.5Write save $17.5Read save $1.4
GPT-6.1 SolAPI ID: gpt-6.1-solLLMFly: ≤272K / >272K · Official: ≤272K / >272K · whole request
≤272K
Input$0.6Output$3Cache write$0.75Cache read$0.03
>272K
Input$1.2Output$4.5Cache write$1.5Cache read$0.06
≤272K
Input$2Output$10Cache write$2.5Cache read$0.1
>272K
Input$4Output$15Cache write$5Cache read$0.2
70% OFF
≤272KIn save $1.4Out save $7Write save $1.75Read save $0.07
>272KIn save $2.8Out save $10.5Write save $3.5Read save $0.14
GPT-6 SolAPI ID: gpt-6-solLLMFly: ≤272K / >272K · Official: ≤272K / >272K · whole request
≤272K
Input$0.6Output$3Cache write$0.75Cache read$0.06
>272K
Input$1.2Output$4.5Cache write$1.5Cache read$0.12
≤272K
Input$2Output$10Cache write$2.5Cache read$0.2
>272K
Input$4Output$15Cache write$5Cache read$0.4
70% OFF
≤272KIn save $1.4Out save $7Write save $1.75Read save $0.14
>272KIn save $2.8Out save $10.5Write save $3.5Read save $0.28
GPT-5.6 SolAPI ID: gpt-5.6-solLLMFly: ≤272K / >272K · Official: ≤272K / >272K · whole request
≤272K
Input$1.5Output$9Cache write$1.875Cache read$0.15
>272K
Input$3Output$13.5Cache write$3.75Cache read$0.3
≤272K
Input$5Output$30Cache write$6.25Cache read$0.5
>272K
Input$10Output$45Cache write$12.5Cache read$1
70% OFF
≤272KIn save $3.5Out save $21Write save $4.375Read save $0.35
>272KIn save $7Out save $31.5Write save $8.75Read save $0.7
GPT-5.6 TerraAPI ID: gpt-5.6-terraLLMFly: ≤272K / >272K · Official: ≤272K / >272K · whole request
≤272K
Input$0.6Output$3.6Cache write$0.75Cache read$0.06
>272K
Input$1.2Output$5.4Cache write$1.5Cache read$0.12
≤272K
Input$2Output$12Cache write$2.5Cache read$0.2
>272K
Input$4Output$18Cache write$5Cache read$0.4
70% OFF
≤272KIn save $1.4Out save $8.4Write save $1.75Read save $0.14
>272KIn save $2.8Out save $12.6Write save $3.5Read save $0.28
GPT-5.5API ID: gpt-5.5Standard
Input$1.5Output$9Cache write—Cache read$0.15
Input$5Output$30Cache write—Cache read$0.5
70% OFF
In save $3.5Out save $21Read save $0.35
Codex Auto ReviewAPI ID: codex-auto-reviewLLMFly: ≤272K / >272K · Official: ≤272K / >272K · whole request
≤272K
Input$0.06Output$0.36Cache write$0.075Cache read$0.006
>272K
Input$0.12Output$0.54Cache write$0.15Cache read$0.012
≤272K
Input$0.2Output$1.2Cache write$0.25Cache read$0.02
>272K
Input$0.4Output$1.8Cache write$0.5Cache read$0.04
70% OFF
≤272KIn save $0.14Out save $0.84Write save $0.175Read save $0.014
>272KIn save $0.28Out save $1.26Write save $0.35Read save $0.028

YOUR PROVIDER → LLMFly

Same model, same usage. What would LLMFly cost?

See your original cost, LLMFly cost and savings side by side. Rates are already filled in—just adjust your usage.

Shared usageDefault: 10M input + 1M output tokens

Azure OpenAI

Original platform$80.00
With LLMFly$24.00
Estimated savings $56.00−70%
Per million tokensInput / outputAzure OpenAI$5.00 / $30.00LLMFly$1.50 / $9.00

LLMFly group: ChatGPT(Codex) · Use the clients supported by this group

Sources and pricing scope

Global Standard · East US 2 · ShortCo · 2026-10-05

Plain text input/output inference only. LLMFly includes the group multiplier. Excludes caching, tools, taxes, top-up fees and negotiated discounts. Matching models does not imply identical APIs or service guarantees.

Original provider pricing ↗LLMFly model rates ↗
Full pricing details →

Amazon Bedrock

Original platform$15.00
With LLMFly$7.50
Estimated savings $7.50−50%
Per million tokensInput / outputAmazon Bedrock$1.00 / $5.00LLMFly$0.50 / $2.50

LLMFly group: Claude Code · Use the clients supported by this group

Sources and pricing scope

Standard · Global Cross Region · ≤200K · 2026-10-05

Plain text input/output inference only. LLMFly includes the group multiplier. Excludes caching, tools, taxes, top-up fees and negotiated discounts. Matching models does not imply identical APIs or service guarantees.

Original provider pricing ↗LLMFly model rates ↗
Full pricing details →

OpenRouter

Original platform$80.00
With LLMFly$24.00
Estimated savings $56.00−70%
Per million tokensInput / outputOpenRouter$5.00 / $30.00LLMFly$1.50 / $9.00

LLMFly group: ChatGPT(Codex) · Use the clients supported by this group

Sources and pricing scope

Standard · ≤272K · 2026-10-05

Plain text input/output inference only. LLMFly includes the group multiplier. Excludes caching, tools, taxes, top-up fees and negotiated discounts. Matching models does not imply identical APIs or service guarantees.

Original provider pricing ↗LLMFly model rates ↗
Full pricing details →

Estimates show USD inference costs for the same model within each card. Models may differ between cards. Codex / Claude Code groups require supported clients.

Choose a model and get started →
02

API pricing calculators

Choose a provider and estimate cost from a specific model and real token usage.

OpenAIOpenAI API pricing calculatorCalculate input, output, and cache chargesAnthropicClaude API pricing calculatorCalculate input, output, and cache chargesGoogleGemini API pricing calculatorCalculate input, output, and cache chargesxAIGrok API pricing calculatorCalculate input, output, and cache charges
03

Do not stop at token price

The number that matters when buying credits

For LLMFly AI credits, use the current price shown in Model Plaza—not the provider reference table above. Then send a small request and check the usage record before projecting monthly spend.

A usable monthly estimate

request.example
monthly_cost =
  (input_tokens / 1_000_000 * input_rate) +
  (output_tokens / 1_000_000 * output_rate) +
  cache_and_tool_charges + retry_cost

Costs teams often miss

  • Conversation history resent on every turn.
  • Reasoning tokens counted as output.
  • AI agent tool loops and validation calls.
  • Retries after 429 or transient upstream failures.
  • Cache writes, cache reads, and storage when the model supports them.

Where to go next

Models and providers

OpenAIGPT-6 Astra; GPT-5.6 Sol, Terra and LunaAnthropicCurrent Claude model IDs and lifecycleGoogleGemini 3.6, 3.7 and 3.8 FlashxAIGrok 4.6

Frequently asked questions

How much does the GPT-6 Astra API cost?

GPT-6 Astra API prices on LLMFly AI are in USD per 1M tokens. For 0–272,000 input tokens per request: input $3, output $15, and cache reads $0.3. For input above 272,000 tokens per request: input $6, output $22.5, and cache reads $0.6. These rates include the 0.3x group multiplier. Cache writes, tools, and other charges are separate; unlisted rates do not mean free usage.

How much does the GPT-5.6 Sol API cost?

GPT-5.6 Sol API prices on LLMFly AI are in USD per 1M tokens. For 0–272,000 input tokens per request: input $1.5, output $9, and cache reads $0.15. For input above 272,000 tokens per request: input $3, output $13.5, and cache reads $0.3. These rates include the 0.3x group multiplier. Cache writes, tools, and other charges are separate; unlisted rates do not mean free usage.

How much does the Claude Fable 5.1 API cost?

Claude Fable 5.1 API prices on LLMFly AI are in USD per 1M tokens. Single platform rate: input $5, output $25, and cache reads $0.125. These rates include the 0.5x group multiplier. Cache writes, tools, and other charges are separate; unlisted rates do not mean free usage.

How much does the Claude Haiku 4.5 API cost?

Claude Haiku 4.5 API prices on LLMFly AI are in USD per 1M tokens. Single platform rate: input $0.5, output $2.5, and cache reads $0.05. These rates include the 0.5x group multiplier. Cache writes, tools, and other charges are separate; unlisted rates do not mean free usage.

How much does the Gemini 3.8 Flash API cost?

Gemini 3.8 Flash API prices on LLMFly AI are in USD per 1M tokens. Single platform rate: input $0.3, output $1.5, and cache reads $0.03. These rates include the 0.4x group multiplier. Cache writes, tools, and other charges are separate; unlisted rates do not mean free usage.

How much does the Grok 4.6 API cost?

Grok 4.6 API prices on LLMFly AI are in USD per 1M tokens. For 0–199,999 input tokens per request: input $0.8, output $2.4, and cache reads $0.2. For input above 199,999 tokens per request: input $1.6, output $4.8, and cache reads $0.4. These rates include the 0.4x group multiplier. Cache writes, tools, and other charges are separate; unlisted rates do not mean free usage.

How much does one LLM API request cost?

Token cost = uncached input tokens × input price per million ÷ 1,000,000 + output tokens × output price per million ÷ 1,000,000, plus cache, tools, and other charges. Use a provider pricing calculator above, select your model, and enter input and output token counts to estimate a request.

How much do API cache reads and cache writes cost?

Cache prices depend on the exact model. The pricing table lists cache reads, cache writes, and applicable 1-hour write rates per million tokens. Reads and writes are separate billing categories; a missing rate or dash does not mean free usage.

Should I apply the discount multiplier to LLMFly AI prices again?

No. The LLMFly AI column and FAQ prices already include the group multiplier. Provider prices are shown separately for comparison. Actual charges depend on your account's applicable rates and usage records.

Check the current model price

Use Model Plaza for the platform price, then verify one request in the usage log.

Open Model Plaza