TechLogHub Data · verified November 8, 2026
LLM API Pricing Compared
What the OpenAI API, Anthropic Claude, Google Gemini, Mistral and xAI Grok charge per million tokens for their current flagship, mid-size and small models, plus OpenAI speech-to-text per minute. Every figure was read on the vendor's own pricing page and links back to it.
Price per 1M tokens (USD)
Standard / on-demand tier, short-context rates. Grouped by provider, ordered flagship → small.
| Provider | Model | Tier | Input | Cached input | Output | Context | Source |
|---|---|---|---|---|---|---|---|
| OpenAI | gpt-6-astra Most capable OpenAI model. Prompt-cache writes billed separately ($12.50 30-min, $25 12-hour per 1M). | Flagship | $10.00 | $1.00 | $50.00 | 1.05M | Pricing |
| OpenAI | gpt-6.1-sol Near-flagship performance at lower cost. | Mid | $2.00 | $0.10 | $10.00 | 1.05M | Pricing |
| OpenAI | gpt-6.1-terra Fast, efficient general model. | Mid | $1.00 | $0.05 | $5.00 | 1.05M | Pricing |
| OpenAI | gpt-6-luna Most efficient OpenAI model for high-volume tasks. | Small | $0.10 | $0.01 | $0.50 | 1.05M | Pricing |
| Anthropic | Claude Fable 5.1 For demanding reasoning and long-horizon agentic work. Full 1M context at standard pricing. | Flagship | $10.00 | $0.25 | $50.00 | 1M | Pricing |
| Anthropic | Claude Opus 5.5 Anthropic's recommended starting model for most workloads. Full 1M context at standard pricing. | Flagship | $4.00 | $0.20 | $20.00 | 1M | Pricing |
| Anthropic | Claude Sonnet 5.5 Balance of speed and intelligence. | Mid | $2.00 | $0.10 | $10.00 | 1M | Pricing |
| Anthropic | Claude Haiku 5.5 Prices shown for prompts up to 100K tokens; prompts over 100K cost $0.50 in / $2.50 out. | Small | $0.10 | $0.01 | $0.50 | 1M | Pricing |
gemini-3.1-pro-preview Preview model. Prices for prompts up to 200K tokens; over 200K: $4 in / $18 out / $0.40 cached. | Flagship | $2.00 | $0.20 | $12.00 | See model docs | Pricing | |
gemini-3.5-flash Google's model page marks this ID deprecated and routed to gemini-3.6-flash; check the models page before building on it. | Mid | $1.50 | $0.15 | $9.00 | 1,048,576 | Pricing | |
gemini-3.1-flash-lite Text/image/video input price; audio input is $0.50. | Small | $0.25 | $0.025 | $1.50 | See model docs | Pricing | |
| Mistral | Mistral Large Current version listed in the docs is Mistral Large 4 (open weights). Batch is 50% off; cached input up to 90% off per the pricing page. | Flagship | $0.50 | — | $1.50 | See model docs | Pricing |
| xAI | grok-4.7 Requests of 200K+ tokens bill all tokens at $4 in / $12 out / $1 cached. | Flagship | $2.00 | $0.50 | $6.00 | 500K | Pricing |
| xAI | grok-4.3 Requests of 200K+ tokens: $2.50 in / $5 out / $0.40 cached. | Mid | $1.25 | $0.20 | $2.50 | 1M | Pricing |
| xAI | grok-build-0.1 Requests of 200K+ tokens: $2 in / $4 out / $0.40 cached. | Small | $1.00 | $0.20 | $2.00 | 256K | Pricing |
Monthly cost calculator
Enter your average tokens per request and monthly volume. Results are sorted cheapest first.
Standard-tier list prices, short-context rates. Excludes batch discounts, cache-write fees, long-context surcharges, taxes and tool/search fees. Models without a published cached rate use the full input price.
| Model | Per request | Per month | |
|---|---|---|---|
| OpenAI gpt-6-luna | $0.00045 | $45 | |
| Anthropic Claude Haiku 5.5 | $0.00045 | $45 | |
| Google gemini-3.1-flash-lite | $0.00125 | $125 | |
| Mistral Mistral Large | $0.00175 | $175 | |
| xAI grok-build-0.1 | $0.00300 | $300 | |
| xAI grok-4.3 | $0.00375 | $375 | |
| OpenAI gpt-6.1-terra | $0.00450 | $450 | |
| xAI grok-4.7 | $0.00700 | $700 | |
| Google gemini-3.5-flash | $0.00750 | $750 | |
| OpenAI gpt-6.1-sol | $0.00900 | $900 | |
| Anthropic Claude Sonnet 5.5 | $0.00900 | $900 | |
| Google gemini-3.1-pro-preview | $0.01000 | $1,000 | |
| Anthropic Claude Opus 5.5 | $0.01800 | $1,800 | |
| OpenAI gpt-6-astra | $0.04500 | $4,500 | |
| Anthropic Claude Fable 5.1 | $0.04500 | $4,500 |
OpenAI Whisper and speech-to-text pricing
OpenAI's estimated cost per audio minute from the transcription section of its pricing page. The legacy whisper-1 model was not on the price list we read; gpt-realtime-whisper is the Whisper-branded live model.
| Model | Use case | Per minute | Per hour | Notes |
|---|---|---|---|---|
| gpt-transcribe | File transcription | $0.0045 | $0.27 | High-accuracy speech-to-text for file and Realtime input. |
| gpt-4o-transcribe | File transcription | $0.006 | $0.36 | Token-billed ($2.50 in / $10 out per 1M); per-minute figure is OpenAI's estimate. |
| gpt-4o-mini-transcribe | File transcription | $0.003 | $0.18 | Token-billed ($1.25 in / $5 out per 1M); per-minute figure is OpenAI's estimate. |
| gpt-realtime-whisper | Live transcription | $0.017 | $1.02 | Whisper-branded live transcription model on the current price list. |
| gpt-live-transcribe | Live transcription | $0.017 | $1.02 | Live transcription. |
DeepSeek (priced in CNY)
DeepSeek's English pricing page did not render its table for us, so these are the peak-hour CNY prices from its Chinese docs page, per 1M tokens. Off-peak prices are half. We do not convert currencies; they are left out of the USD calculator.
Get pricing changes in your inbox. One short digest a week: new tools, price changes and the comparisons developers actually read.
Get the TechLogHub digest
One email a week: new and trending developer tools, fresh comparisons, and what shipped. Unsubscribe in one click.
Methodology
On November 8, 2026we opened each vendor's official API pricing page and recorded the Standard (on-demand) list price per 1M input, cached-input and output tokens for the current flagship, mid-size and small text models. Context windows come from the vendor's model documentation; where we could not confirm one it says "See model docs". Where a vendor prices by prompt length (Gemini 3.1 Pro over 200K tokens, xAI at 200K+, Claude Haiku 5.5 over 100K), the table shows the short-prompt rate and the note gives the long-prompt rate. Mistral appears with the single model its public pricing page prices; DeepSeek appears in CNY only. Batch, flex and priority tiers, fine-tuning, image, embeddings and tool fees are excluded. No figures come from third-party aggregators.
LLM API pricing: FAQ
Quick answers to common questions
How much does the OpenAI API cost per 1M tokens?
On 2026-11-08, OpenAI's Standard-tier list prices per 1M tokens were: gpt-6-astra $10 input / $50 output, gpt-6.1-sol $2 / $10, gpt-6.1-terra $1 / $5 and gpt-6-luna $0.10 / $0.50. Cached input is billed at a lower rate. Check developers.openai.com/api/docs/pricing for current numbers.
How much does the ChatGPT API cost for a typical app?
Multiply tokens per request by requests per month. For example, 100,000 requests a month at 2,000 input and 500 output tokens is 200M input and 50M output tokens: on gpt-6.1-terra ($1 / $5 per 1M) that is $200 + $250 = $450 a month before caching or batch discounts. Use the calculator on this page with your own numbers.
How much does OpenAI Whisper / speech-to-text cost?
OpenAI's current price list shows gpt-transcribe at an estimated $0.0045 per minute, gpt-4o-transcribe at $0.006 per minute and gpt-4o-mini-transcribe at $0.003 per minute for file transcription. The Whisper-branded live model, gpt-realtime-whisper, is listed at $0.017 per minute.
Which LLM API is cheapest?
Among the models in this table, the lowest list prices are the small tiers: gpt-6-luna and Claude Haiku 5.5 (under 100K-token prompts) at $0.10 input / $0.50 output per 1M tokens, followed by gemini-3.1-flash-lite at $0.25 / $1.50. DeepSeek publishes low CNY prices. The cheapest model for you depends on quality: test on your own prompts before switching.
Why do my real API bills differ from list prices?
List prices exclude several things: output tokens from reasoning, cache-write fees, long-context surcharges (Gemini 3.1 Pro over 200K tokens, xAI at 200K+, Claude Haiku 5.5 over 100K), regional or data-residency uplifts, tool calls such as web search, and taxes. Batch APIs and prompt caching can reduce costs substantially.