Cost of 1M input tokens right now: Claude vs GPT vs Gemini
32 Anthropic, OpenAI and Google models we track, priced in USD per million tokens from each vendor’s official pricing page. The cheapest input today is Claude Haiku 5.5 at $0.10; the most expensive is Claude Fable 5.1 at $10.00 (about 100× more).
Prices checked: Anthropic Claude 2026-10-11 · OpenAI GPT 2026-10-11 · Google Gemini 2026-10-11. This page is rebuilt with every price sync.
Anthropic Claude
| Model | 1M input | Cached input | 1M output | Long-context input | 1M in + 100K out | Context |
|---|---|---|---|---|---|---|
| Claude Haiku 5.5 | $0.10 | $0.010 | $0.50 | $0.50 (>100K) | $0.15 | 1M |
| Claude Haiku 4.5 | $1.00 | $0.10 | $5.00 | — | $1.50 | 200K |
| Claude Sonnet 5 | $2.00 | $0.20 | $10.00 | — | $3.00 | 1M |
| Claude Sonnet 5.5 | $2.00 | $0.10 | $10.00 | — | $3.00 | 1M |
| Claude Sonnet 4.6 | $3.00 | $0.30 | $15.00 | — | $4.50 | 1M |
| Claude Opus 5.5 | $4.00 | $0.20 | $20.00 | — | $6.00 | 1M |
| Claude Opus 4.8 | $5.00 | $0.50 | $25.00 | — | $7.50 | 1M |
| Claude Opus 5 | $5.00 | $0.50 | $25.00 | — | $7.50 | 1M |
| Claude Opus 4.7 | $5.00 | $0.50 | $25.00 | — | $7.50 | 1M |
| Claude Fable 5.1 | $10.00 | $0.25 | $50.00 | — | $15.00 | 1M |
| Claude Fable 5 | $10.00 | $1.00 | $50.00 | — | $15.00 | 1M |
| Claude Mythos 5.1 | $10.00 | $0.25 | $50.00 | — | $15.00 | 1M |
OpenAI GPT
| Model | 1M input | Cached input | 1M output | Long-context input | 1M in + 100K out | Context |
|---|---|---|---|---|---|---|
| GPT-6 Luna | $0.10 | $0.010 | $0.50 | $0.20 (>272K) | $0.15 | 1.1M |
| GPT-5.6 Luna | $0.20 | $0.020 | $1.20 | $0.40 (>272K) | $0.32 | 1.1M |
| GPT-5.1 | $1.25 | $0.13 | $10.00 | — | $2.25 | 400K |
| GPT-5.6 Terra | $2.00 | $0.20 | $12.00 | $4.00 (>272K) | $3.20 | 1.1M |
| GPT-6.1 Sol | $2.00 | $0.10 | $10.00 | $4.00 (>272K) | $3.00 | 1.1M |
| GPT-6 Sol | $2.00 | $0.20 | $10.00 | $4.00 (>272K) | $3.00 | 1.1M |
| GPT-5.4 | $2.50 | $0.25 | $15.00 | $5.00 (>272K) | $4.00 | 400K |
| GPT-5.6 Sol | $4.00 | $0.40 | $20.00 | $8.00 (>272K) | $6.00 | 1.1M |
| GPT-5.5 | $5.00 | $0.50 | $30.00 | $10.00 (>272K) | $8.00 | 400K |
| GPT-6 Astra | $10.00 | $1.00 | $50.00 | $20.00 (>272K) | $15.00 | 1.1M |
Google Gemini
| Model | 1M input | Cached input | 1M output | Long-context input | 1M in + 100K out | Context |
|---|---|---|---|---|---|---|
| Gemini 2.5 Flash-Lite | $0.10 | $0.010 | $0.40 | — | $0.14 | 1M |
| Gemini 3.1 Flash-Lite | $0.25 | $0.025 | $1.50 | — | $0.40 | 1M |
| Gemini 2.5 Flash | $0.30 | $0.030 | $2.50 | — | $0.55 | 1M |
| Gemini 3.5 Flash-Lite | $0.30 | $0.030 | $2.50 | — | $0.55 | 1.0M |
| Gemini 3.7 Flash | $0.75 | $0.075 | $3.75 | — | $1.13 | 1.0M |
| Gemini 3.6 Flash | $0.75 | $0.075 | $3.75 | — | $1.13 | 1M |
| Gemini 3.8 Flash | $0.75 | $0.075 | $3.75 | — | $1.13 | 1.0M |
| Gemini 2.5 Pro | $1.25 | $0.13 | $10.00 | $2.50 (>200K) | $2.25 | 2M |
| Gemini 3.5 Flash | $1.50 | $0.15 | $9.00 | — | $2.40 | 1M |
| Gemini 3.1 Pro Preview | $2.00 | $0.20 | $12.00 | $4.00 (>200K) | $3.20 | 2M |
How to read this
“1M input” is the standard rate for non-cached prompt tokens on short prompts. If one request is long enough to cross a vendor’s long-context threshold, the whole request is billed at the “Long-context input” rate instead, so sending 1M tokens as one huge prompt can cost more than the same tokens split across smaller requests.
Batch APIs, regional endpoints and enterprise discounts are not included. For your own mix of input, output and cache hits, use the cost estimator.
FAQ
How much do 1 million input tokens cost on Claude, GPT and Gemini right now?
At standard (non-cached, short-prompt) rates: the cheapest Anthropic Claude model is Claude Haiku 5.5 at $0.10; the cheapest OpenAI GPT model is GPT-6 Luna at $0.10; the cheapest Google Gemini model is Gemini 2.5 Flash-Lite at $0.10. Across all 32 models in this table the range is $0.10 (Claude Haiku 5.5) to $10.00 (Claude Fable 5.1) per 1M input tokens. Prices were last checked against official pricing pages between 2026-10-11 and 2026-10-11.
Does sending 1M tokens in one request cost more than sending them in many small requests?
For some models, yes. Models with a long-context tier bill the whole request at a higher rate once a single prompt crosses the tier threshold. 12 of the 32 models here have such a tier; the "Long-context input" column shows the rate and where it starts. Splitting the work into smaller requests keeps the standard rate.
How much cheaper are cached input tokens?
Where a vendor publishes a cached-input price it is listed in the "Cached input" column, for example Claude Haiku 5.5: $0.010 cached vs $0.10 standard; Claude Haiku 4.5: $0.10 cached vs $1.00 standard; Claude Sonnet 5: $0.20 cached vs $2.00 standard. Some vendors charge extra for cache writes, so the real saving depends on your hit rate.
What does a job with 1M input and 100K output tokens cost?
See the "1M in + 100K out" column: it ranges from $0.14 to $15.00 across these models at standard rates. Use the estimator at https://www.llmabacus.com/estimate for your own token counts.