LLM API Pricing Tracker: OpenAI, GPT-6 and Claude Prices
Updated 22 September 2026 · first published 4 July 2026
OpenAI API pricing and Claude API pricing in one normalised table: input, output, cache read/write, context windows and batch discounts per 1 million tokens. The OpenAI and Anthropic rows are refreshed live from each provider's own pricing page, so GPT pricing and Claude pricing here match what you are billed, and new models show up as soon as they are listed.
This is API pricing only. ChatGPT and Codex subscriptions are separate monthly plans; see the FAQ below for how they relate.
- All prices in USD per 1 million tokens
- Estimated values shown in gray
- Cache prices reflect prompt caching where available
Methodology
Collection
Pricing data collected from official provider pricing pages and API documentation.
Normalisation
All prices normalised to USD per 1 million tokens for direct comparison.
Cache pricing
Cache Read/Write columns show prompt caching discounts where providers support them.
Display policy
Official values and estimated values are clearly separated. Estimates marked in gray.
Update Feed
GPT-6 and Claude Opus 5.5 added; models now discovered automatically
Added GPT-6-Astra ($10/$50), GPT-6-Sol ($2/$10), GPT-6-Luna ($0.10/$0.50) and Claude Opus 5.5 ($4/$20). Dropped GPT-5.6-Terra and GPT-5.6-Luna, which OpenAI no longer lists. Any model OpenAI or Anthropic adds to its pricing page now appears in the table automatically.
LLM API Pricing Tracker launched
Published per-token pricing comparison across 12+ major LLM providers as a FinOps LLM research article.
Per-Token Pricing Table
| Provider ↕ | Model ↕ | Input $/1M ↕ | Output $/1M ↕ | Cache Read $/1M ↕ | Cache Write $/1M ↕ | Context Window ↕ | Batch Discount |
|---|---|---|---|---|---|---|---|
| OpenAI | GPT-6-Astra | $10.00 | $50.00 | $1.00 | $12.50 | — | 50% |
| OpenAI | GPT-6-Sol | $2.00 | $10.00 | $0.20 | $2.50 | — | 50% |
| OpenAI | GPT-6-Luna | $0.10 | $0.50 | $0.01 | $0.125 | — | 50% |
| OpenAI | GPT-5.6-Sol | $4.00 | $20.00 | $0.40 | $5.00 | 1.05M | 50% |
| Anthropic | Claude Fable 5.1 | $10.00 | $50.00 | $0.25 | $12.50 | 1M | 50% |
| Anthropic | Claude Opus 5.5 | $4.00 | $20.00 | $0.20 | $5.00 | 1M | 50% |
| Anthropic | Claude Fable 5 | $10.00 | $50.00 | $1.00 | $12.50 | 1M | 50% |
| Anthropic | Claude Opus 5 | $5.00 | $25.00 | $0.50 | $6.25 | 1M | 50% |
| Anthropic | Claude Sonnet 5* | $2.00 | $10.00 | $0.20 | $2.50 | 1M | 50% |
| Anthropic | Claude Sonnet 4.6 | $3.00 | $15.00 | $0.30 | $3.75 | 1M | 50% |
| Anthropic | Claude Haiku 4.5 | $1.00 | $5.00 | $0.10 | $1.25 | 200K | 50% |
| Gemini 3.8 Flash | $0.75 | $3.75 | $0.075 | — | — | 50% | |
| Gemini 3.7 Flash | $0.75 | $3.75 | $0.075 | — | — | 50% | |
| Gemini 3.6 Flash | $0.75 | $3.75 | $0.075 | — | — | 50% | |
| Gemini 3.5 Flash | $1.50 | $9.00 | $0.15 | — | — | 50% | |
| Gemini 3.5 Flash-Lite | $0.30 | $2.50 | $0.03 | — | — | 50% | |
| Gemini 3.1 Flash-Lite | $0.25 | $1.50 | $0.025 | — | — | 50% | |
| Gemini 3.1 Pro Preview | $2.00 | $12.00 | $0.20 | — | — | 50% | |
| Gemini 3 Flash Preview | $0.50 | $3.00 | $0.05 | — | — | 50% | |
| Gemini 2.5 Pro | $1.25 | $10.00 | $0.125 | — | — | 50% | |
| Gemini 2.5 Flash | $0.30 | $2.50 | $0.03 | — | — | 50% | |
| Gemini 2.5 Flash-Lite | $0.10 | $0.40 | $0.01 | — | — | 50% | |
| xAI | grok-4.7 | $2.00 | $6.00 | $0.50 | — | 500K | — |
| xAI | grok-build-0.1 | $1.00 | $2.00 | $0.20 | — | 256K | — |
| xAI | grok-4.6 | $2.00 | $6.00 | $0.50 | — | 500K | — |
| xAI | grok-4.5 | $2.00 | $6.00 | $0.30 | — | 500K | — |
| xAI | grok-4.3 | $1.25 | $2.50 | $0.20 | — | 1M | — |
| xAI | grok-4.20-multi-agent-0309 | $1.25 | $2.50 | $0.20 | — | 1M | — |
| xAI | grok-4.20-0309-reasoning | $1.25 | $2.50 | $0.20 | — | 1M | — |
| xAI | grok-4.20-0309-non-reasoning | $1.25 | $2.50 | $0.20 | — | 1M | — |
| Meta | Llama 4 Maverick (DeepInfra) | $0.20 | $0.80 | - | - | 1M | - |
| Meta | Llama 4 Scout (Groq) | $0.11 | $0.34 | - | - | 5M | - |
| Meta | Llama 3.3 70B (Groq) | $0.59 | $0.79 | - | - | 8M | - |
| Mistral | Mistral Large 2/3 | $2.00 | $6.00 | - | - | 128K | - |
| Mistral | Mistral Medium 3.5 | $1.50 | $7.50 | - | - | 128K | - |
| Mistral | Mistral Small 3 | $0.10 | $0.30 | - | - | 128K | - |
| Mistral | Codestral | $0.30 | $0.90 | - | - | 128K | - |
| DeepSeek | deepseek-flash | $0.30 | $1.20 | $0.006 | — | 1M | — |
| DeepSeek | deepseek-v4-pro | $1.32 | $3.96 | $0.044 | — | 1M | — |
| Alibaba Qwen | Qwen 3.7-Max | $2.50 | $7.50 | - | - | 131K | - |
| Moonshot Kimi | Kimi K2.6 | $0.95 | $4.00 | $0.16 | - | 200K | - |
| Xiaomi MiMO Sign up - both get $2 ↗ |
MiMo-V2-Pro | $0.50 | $2.00 | - | - | 128K | - |
| Xiaomi MiMO Sign up - both get $2 ↗ |
MiMo-V2-Omni | $0.40 | $1.60 | - | - | 128K | - |
| Zhipu AI | GLM-5 | $1.00 | $4.00 | - | - | 128K | - |
| Zhipu AI | GLM-5-Turbo | $0.30 | $1.20 | - | - | 128K | - |
| MiniMax | M2.7 | $0.40 | $1.60 | - | - | 128K | - |
| StepFun | step-3.5-flash | $0.30 | $1.20 | - | - | 128K | - |
* Claude Sonnet 5 is $2/$10 per 1M tokens permanently — the increase to $3/$15 that had been scheduled for 1 September 2026 was cancelled. Cache-write figures are the 5-minute TTL; the 1-hour TTL costs 2× the base input rate. Claude Opus 4.8, 4.7 and 4.6 share Claude Opus 5 pricing. Claude 4.7 and newer use a tokenizer that produces roughly 30% more tokens for the same text, so compare cost per task, not cost per token.
Values not explicitly provided by providers are shown in gray.
Meta Llama prices reflect hosted API pricing (Fireworks, Together, etc.), not self-hosted cost. Self-hosting economics are covered in our Open-Source vs Closed-Source Cost Comparison.
Context windows are maximum supported; actual availability may vary by tier.
Cache Read/Write pricing reflects provider-specific prompt caching where available. "-" means the provider does not offer caching or pricing is not published.
Batch Discount shows the approximate savings when using batch/async APIs where available.
Some links may include referral codes.
Prices can change frequently. Always verify the latest information on official provider pages.
Practical Guide
How to use this pricing data for cost planning
1) Calculate your token volume first
Before comparing prices, estimate your monthly input and output token volumes. Output tokens are typically 3-5x more expensive than input, so reducing output length has the biggest impact on cost.
2) Factor in caching savings
If your workload has repeated system prompts or context (RAG, multi-turn conversations), prompt caching can reduce input costs by 50-90%. Check the Cache Read column for applicable discounts.
3) Use batch APIs for non-urgent work
Most major providers offer 50% discounts on batch/async API calls. If your workload tolerates latency (evaluations, data processing, content generation), batch pricing cuts costs in half.
4) Compare total cost, not just per-token price
A cheap model that needs 3x more tokens to match quality may cost more overall. Factor in model quality and task-specific performance alongside raw pricing.
Next step: Compare your spend across providers—see Azure OpenAI vs Direct OpenAI Cost for a breakdown of managed vs direct pricing models.
Want this applied to your own LLM spend? FinOps LLM runs a free audit of your AI costs and shows where the savings are. Book free audit →
FAQ
What is the current OpenAI API pricing?
Per 1 million tokens at standard rates: GPT-6-Astra costs $10 input and $50 output, GPT-6-Sol $2 input and $10 output, GPT-6-Luna $0.10 input and $0.50 output, and GPT-5.6-Sol $4 input and $20 output. Cached input is billed at a tenth of the input price, and the Batch API is 50% off.
How much does GPT-6 cost?
GPT-6 comes in three tiers. Astra is $10 per million input tokens and $50 per million output tokens, Sol is $2 and $10, and Luna is $0.10 and $0.50. Above the long-context threshold Astra rises to $20 input and $75 output, and Sol to $4 and $15.
What is Claude API pricing?
Per 1 million tokens: Claude Fable 5.1 is $10 input and $50 output, Claude Opus 5.5 is $4 and $20, Claude Sonnet 5 is $2 and $10, and Claude Haiku 4.5 is $1 and $5. Prompt-cache writes cost 1.25x input for a 5-minute cache, and the Message Batches API is 50% off.
How much does Claude Opus 5.5 cost?
Claude Opus 5.5 is $4 per million input tokens and $20 per million output tokens, with cache hits at $0.20 and 5-minute cache writes at $5. Batch processing halves that to $2 and $10. Fast mode costs $8 input and $40 output.
Is ChatGPT or Codex pricing the same as OpenAI API pricing?
No. ChatGPT and Codex subscriptions are flat monthly plans with usage limits, billed separately from the API. The OpenAI API bills per token on your platform account, at the rates in this table, whether the calls come from your own app or from Codex signed in with an API key.
How often are these prices updated?
The OpenAI and Anthropic rows are read from each provider's published pricing page and are never more than six hours old. New models either provider lists are added automatically. If a page fails our sanity checks, the table falls back to the last verified prices rather than showing a guess.