AI model comparison
Current Claude, GPT, Gemini, Grok and DeepSeek models side by side: what they cost through the API and how much text they can take in and write out.
All models
| Model | Input / M | Cached input / M | Output / M | Batch | Context | Max output | Cutoff or update | Notes |
|---|---|---|---|---|---|---|---|---|
| Claude Fable 5.1 Anthropic | $10.00 | $0.25 | $50.00 | 50% | 1M | 128K | Jun 2026 (knowledge cutoff) | Prices · Limits |
| Claude Opus 5.5 Anthropic | $4.00 | $0.20 | $20.00 | 50% | 1M | 128K | Jun 2026 (knowledge cutoff) | Prices · Limits |
| Claude Sonnet 5.5 Anthropic | $2.00 | $0.10 | $10.00 | 50% | 1M | 128K | Jun 2026 (knowledge cutoff) | Prices · Limits |
| Claude Haiku 5.5 Anthropic | $0.10 | $0.010 | $0.50 | 50% | 1M | 128K | Jun 2026 (knowledge cutoff) | Prompts over 100K tokens cost $0.50 in and $2.50 out. Prices · Limits |
| GPT-6 Astra OpenAI | $10.00 | $1.00 | $50.00 | 50% | 1.05M | 128K | Apr 30, 2026 (knowledge cutoff) | Prompts over 272K tokens cost $20.00 in and $75.00 out. Prices · Limits |
| GPT-6.1 Sol OpenAI | $2.00 | $0.10 | $10.00 | 50% | 1.05M | 128K | Apr 30, 2026 (knowledge cutoff) | Prompts over 272K tokens cost $4.00 in and $15.00 out. Prices · Limits |
| GPT-6.1 Terra OpenAI | $1.00 | $0.050 | $5.00 | 50% | 1.05M | 128K | Jun 1, 2026 (knowledge cutoff) | Prompts over 272K tokens cost $2.00 in and $7.50 out. Prices · Limits |
| GPT-6 Luna OpenAI | $0.10 | $0.010 | $0.50 | 50% | 1.05M | 128K | May 18, 2026 (knowledge cutoff) | Prompts over 272K tokens cost $0.20 in and $0.75 out. Prices · Limits |
| Gemini 4 Argon Google | $2.00 | $0.10 | $10.00 | 50% | 1.05M | 66K | Oct 2026 (latest update) | From 2026-12-01: $4.00 in, $20.00 out. Prices · Limits |
| Gemini 3.8 Flash Google | $0.75 | $0.075 | $3.75 | 50% | 1.05M | 66K | Sep 2026 (latest update) | From 2027-01-01: $1.50 in, $7.50 out. Prices · Limits |
| Gemini 3.5 Flash-Lite Google | $0.30 | $0.030 | $2.50 | 50% | 1.05M | 66K | Jul 2026 (latest update) | Prices · Limits |
| Grok 4.7 xAI | $2.00 | $0.50 | $6.00 | 80% | 500K | – | – | Prompts over 200K tokens cost $4.00 in and $12.00 out. Prices · Limits |
| DeepSeek Flash DeepSeek | $0.30 | $0.006 | $1.20 | None | 1M | 384K | – | Peak-hour price. Off-peak hours cost half. Prices · Limits |
| DeepSeek V4 Pro DeepSeek | $1.32 | $0.044 | $3.96 | None | 1M | 384K | – | Peak-hour price. Off-peak hours cost half. Prices · Limits |
How it’s calculated
Prices are standard API prices in U.S. dollars per million tokens, copied from each provider’s pricing page and checked every week; a price change flags the model for review. Where a provider has announced a new price, the table switches on that date. Batch discount is the share of the standard price charged for batch requests.
Context window is the most tokens a request can hold (prompt plus, for some providers, the reply); maximum output is the longest reply. Providers do not list launch dates on their model pages, so the date column shows what they do publish: the knowledge cutoff (Anthropic, OpenAI) or the date of the latest update (Google). A token is roughly three quarters of an English word.
Example
A request with 10,000 input tokens and 1,000 output tokens on a model priced at $2 input and $10 output costs 10,000 × $2 ÷ 1,000,000 + 1,000 × $10 ÷ 1,000,000 = $0.03.