Long-context pricing: where AI API prices rise
7 models in our data charge more once a single request passes a token threshold: GPT-6 Astra, GPT-6 Sol, GPT-6.1 Sol and GPT-6 Luna above 272,000 input tokens; Gemini 3.1 Pro Preview and Grok 4.7 above 200,000 input tokens; Claude Haiku 5.5 above 100,000 input tokens (providers’ own price lists, prices as listed in the United States on ). Above it the input price is 2 to 5 times the standard one, and the whole request, cached input and output included, is billed at the higher rate.
Where the price rises
| Model | Context window, tokens | Higher price | Input, per 1M | Output, per 1M | Cache read, per 1M | Cache write, per 1M | Price list |
|---|---|---|---|---|---|---|---|
| Claude Haiku 5.5Anthropic | 1,000,000 | above 100,000 tokens | $0.10 → $0.50 (×5) | $0.50 → $2.50 (×5) | $0.01 → $0.05 (×5) | $0.125 → $0.625 (×5) | platform.claude.com, read |
| Gemini 3.1 Pro PreviewGoogle | 1,048,576 | above 200,000 tokens | $2.00 → $4.00 (×2) | $12.00 → $18.00 (×1.5) | $0.20 → $0.40 (×2) | UnknownNot listed | ai.google.dev, read |
| Grok 4.7xAI | 500,000 | above 200,000 tokens | $2.00 → $4.00 (×2) | $6.00 → $12.00 (×2) | $0.50 → $1.00 (×2) | UnknownNot listed | docs.x.ai, read |
| GPT-6 AstraOpenAI | 1,050,000 | above 272,000 tokens | $10.00 → $20.00 (×2) | $50.00 → $75.00 (×1.5) | $1.00 → $2.00 (×2) | $12.50 → $25.00 (×2) | developers.openai.com, read |
| GPT-6 SolOpenAI | 1,050,000 | above 272,000 tokens | $2.00 → $4.00 (×2) | $10.00 → $15.00 (×1.5) | $0.20 → $0.40 (×2) | $2.50 → $5.00 (×2) | developers.openai.com, read |
| GPT-6.1 SolOpenAI | 1,050,000 | above 272,000 tokens | $2.00 → $4.00 (×2) | $10.00 → $15.00 (×1.5) | $0.10 → $0.20 (×2) | $2.50 → $5.00 (×2) | developers.openai.com, read |
| GPT-6 LunaOpenAI | 1,050,000 | above 272,000 tokens | $0.10 → $0.20 (×2) | $0.50 → $0.75 (×1.5) | $0.01 → $0.02 (×2) | $0.125 → $0.25 (×2) | developers.openai.com, read |
Published Provider price lists, prices as listed in the United States on . Below → above the threshold; the whole request above it is billed at the higher rate.
Crossing the threshold
The jump is per request, not per token above the line: one request just over the threshold costs this much more than one that stays at it.
| Model | At the threshold | 1,000 tokens above | Difference |
|---|---|---|---|
| Claude Haiku 5.5Anthropic | $0.011 | $0.056 | +404.5 % |
| Gemini 3.1 Pro PreviewGoogle | $0.424 | $0.840 | +98.1 % |
| Grok 4.7xAI | $0.412 | $0.828 | +101.0 % |
| GPT-6 AstraOpenAI | $2.82 | $5.61 | +98.9 % |
| GPT-6 SolOpenAI | $0.564 | $1.12 | +98.9 % |
| GPT-6.1 SolOpenAI | $0.564 | $1.12 | +98.9 % |
| GPT-6 LunaOpenAI | $0.028 | $0.056 | +98.9 % |
Calculated One request, no cache, 2,000 output tokens, at the listed prices above.
What one long request costs
| Model | At the long-context price | At the standard price | Long-context price adds |
|---|---|---|---|
| Claude Haiku 5.5Anthropic | $0.155 | $0.031 | +400.0 % |
| Gemini 3.1 Pro PreviewGoogle | $1.24 | $0.624 | +98.1 % |
| Grok 4.7xAI | $1.22 | $0.612 | +100.0 % |
| GPT-6 AstraOpenAI | $6.15 | $3.10 | +98.4 % |
| GPT-6 SolOpenAI | $1.23 | $0.620 | +98.4 % |
| GPT-6.1 SolOpenAI | $1.23 | $0.620 | +98.4 % |
| GPT-6 LunaOpenAI | $0.061 | $0.031 | +98.4 % |
Calculated One request of 300,000 input and 2,000 output tokens, no cache.
Price lists without a long-context rate in our data
“Claude 4.6 and later models (except Claude Haiku 5.5) and Claude Mythos Preview include the full 1M token context window at standard pricing.”
| Model | Context window, tokens | Long-context price |
|---|---|---|
| GPT-5.4 MiniOpenAI | 400,000 | UnknownNo higher rate in our data |
| Claude Fable 5.1Anthropic | 1,000,000 | UnknownNo higher rate in our data |
| Claude Opus 5.5Anthropic | 1,000,000 | UnknownNo higher rate in our data |
| Claude Sonnet 5.5Anthropic | 1,000,000 | UnknownNo higher rate in our data |
| Gemini 3.8 FlashGoogle | 1,048,576 | UnknownNo higher rate in our data |
| Gemini 4 ArgonGoogle | 1,048,576 | UnknownNo higher rate in our data |
| DeepSeek V4 Pro 0813DeepSeek | 1,048,576 | UnknownNo higher rate in our data |
| DeepSeek V4.1 Flash 1017DeepSeek | 1,048,576 | UnknownNo higher rate in our data |
| Mistral Large 3 2512Mistral | 262,144 | UnknownNo higher rate in our data |
| Mistral Large 4Mistral | 1,048,576 | UnknownNo higher rate in our data |
| Mistral Medium 3.5Mistral | 262,144 | UnknownNo higher rate in our data |
| Qwen3.8 FlashQwen | 1,000,000 | UnknownNo higher rate in our data |
| Qwen3.8 Max (0902)Qwen | 1,000,000 | UnknownNo higher rate in our data |
| Qwen3.8 27BQwen | 1,024,000 | UnknownNo higher rate in our data |
| gpt-oss-120bOpenAI | 131,072 | UnknownNo higher rate in our data |
| gpt-oss-20bOpenAI | 131,072 | UnknownNo higher rate in our data |
| GLM 5.3Z.ai | 1,048,576 | UnknownNo higher rate in our data |
| Kimi K3MoonshotAI | 1,048,576 | UnknownNo higher rate in our data |
| MiniMax M3MiniMax | 1,048,576 | UnknownNo higher rate in our data |
All API prices · Cheapest per task · Caching and batch · Rate limits by tier · What breaks when you switch · Rate limit planner · Task cost calculator · All models on OpenRouter · How we check prices