Claude Sonnet 5 Pricing (2026): API Cost per 1M Tokens in $ and ₹, plus Sonnet 5.5, Opus and Fable
Claude Sonnet 5 costs $2 per 1M input tokens and $10 per 1M output tokens. Its successor, Claude Sonnet 5.5, has exactly the same price. On Velona that is ₹204.53 per 1M input tokens and ₹1,023 per 1M output tokens, paid from a rupee wallet with UPI. Prices are as of 16 October 2026.
This guide gives the price of every current Claude model in dollars and rupees, shows what real requests and real months cost, and helps you pick the cheapest model that does the job.
₹204.53
Sonnet 5 and 5.5, per 1M input tokens on Velona ($2)
₹1,023
Per 1M output tokens ($10). Output costs 5 times input.
₹0.72
One typical chat reply on Sonnet 5.5 (1,000 tokens in, 500 out)
Claude API pricing for every model (per 1M tokens)
Anthropic bills per million tokens, split into input (what you send) and output (what the model writes back). A token is roughly three quarters of an English word. Velona converts the same list price to rupees at the live exchange rate.
| Model and Velona ID | Input $ | Output $ | Cache read $ | Input ₹ on Velona | Output ₹ on Velona |
|---|---|---|---|---|---|
Claude Haiku 5.5 Currentanthropic/claude-haiku-5.5 | $0.10 | $0.50 | $0.01 | ₹10.23 | ₹51.13 |
Claude Sonnet 5.5 Currentanthropic/claude-sonnet-5.5 | $2 | $10 | $0.10 | ₹204.53 | ₹1,023 |
Claude Sonnet 5 Previousanthropic/claude-sonnet-5 | $2 | $10 | $0.20 | ₹204.53 | ₹1,023 |
Claude Opus 5.5 Currentanthropic/claude-opus-5.5 | $4 | $20 | $0.20 | ₹409.06 | ₹2,045 |
Claude Fable 5.1 Currentanthropic/claude-fable-5.1 | $10 | $50 | $0.25 | ₹1,023 | ₹5,113 |
Claude Haiku 4.5 Previousanthropic/claude-haiku-4.5 | $1 | $5 | $0.10 | ₹102.27 | ₹511.33 |
Claude Opus 4.8 Previousanthropic/claude-opus-4.8 | $5 | $25 | $0.50 | ₹511.33 | ₹2,557 |
Claude Fable 5 Previousanthropic/claude-fable-5 | $10 | $50 | $1 | ₹1,023 | ₹5,113 |
Haiku 5.5 prices apply to prompts up to 100K tokens. Longer prompts are billed at $0.50 input and $2.50 output. All current models have a 1M token context window. Rupee prices follow the live USD to INR rate and move slightly through the day.
How the Claude models compare on price
Output tokens drive most bills, so this chart compares the output price. The gap between the cheapest and the most expensive Claude model is 100 times.
Output price
Rupees per 1M output tokens on Velona
Current Claude models. Hover a bar for the exact figure.
Claude Haiku 5.5 · ₹10.23 input
Claude Sonnet 5.5 · ₹204.53 input
Claude Opus 5.5 · ₹409.06 input
Claude Fable 5.1 · ₹1,023 input
Sonnet 5 and Sonnet 5.5 cost the same. Opus 5.5 is twice Sonnet, and Fable 5.1 is five times Sonnet.
What one Claude request really costs in rupees
Per million prices are hard to picture. Pick a task below to see what a single request costs on each current model.
Cost per request
One request, four Claude models
1,000 input and 500 output tokens
Claude Haiku 5.5
Claude Sonnet 5.5
Claude Opus 5.5
Claude Fable 5.1
6,000 input and 600 output tokens
Claude Haiku 5.5
Claude Sonnet 5.5
Claude Opus 5.5
Claude Fable 5.1
30,000 input and 1,000 output tokens
Claude Haiku 5.5
Claude Sonnet 5.5
Claude Opus 5.5
Claude Fable 5.1
15,000 input and 2,000 output tokens
Claude Haiku 5.5
Claude Sonnet 5.5
Claude Opus 5.5
Claude Fable 5.1
No prompt caching assumed, so these are upper bounds. Repeated context, like a system prompt or a codebase an agent rereads, gets much cheaper with caching.
Where your rupees go: input or output?
Chat is output heavy. Agents are input heavy, because they resend the code and history on every step. That tells you which lever to pull.
Cost split
Share of each request spent on input vs output
- Input tokens
- Output tokens
A chat reply · ₹0.72 on Sonnet 5.5
Input 29% · Output 71%
Summarise a 10 page document · ₹1.84 on Sonnet 5.5
Input 67% · Output 33%
One coding agent step · ₹7.16 on Sonnet 5.5
Input 86% · Output 14%
Review a pull request · ₹5.11 on Sonnet 5.5
Input 60% · Output 40%
For chat, cap max_tokens. For agents, trim the context you resend or use caching.
Your monthly Claude bill, three real examples
Monthly cost
What a month costs on each model
25 chat replies a day for 30 days (1,000 input and 500 output tokens each)
Claude Haiku 5.5
Claude Sonnet 5.5
Claude Opus 5.5
Claude Fable 5.1
150 coding agent steps a day for 22 workdays (30,000 input and 1,000 output tokens each)
Claude Haiku 5.5
Claude Sonnet 5.5
Claude Opus 5.5
Claude Fable 5.1
10,000 customer chats a month (2,500 input and 400 output tokens each)
Claude Haiku 5.5
Claude Sonnet 5.5
Claude Opus 5.5
Claude Fable 5.1
A student chatting on Sonnet 5.5 every day spends about ₹537 a month. A developer running a coding agent all day spends about ₹23,623 on Sonnet 5.5, or ₹1,181 on Haiku 5.5.
Which Claude model should you pick?
Start with the cheapest model that can do the job, and move up only when the output is not good enough. This flow covers most cases.
Decision guide
Match the job to the model
Cost shown is one chat reply (1,000 tokens in, 500 out)
A common pattern: send every request to Haiku 5.5 first, and route only the hard ones to Sonnet 5.5 or Opus 5.5. On Velona all of them sit behind one API key, so switching is a one word change.
Claude Sonnet 5 vs Sonnet 5.5: is there a price difference?
Claude Sonnet 5
$2 / $10₹204.53 input and ₹1,023 output per 1M on Velona. Cache reads cost $0.20 per 1M. Now the previous generation, still available.
Claude Sonnet 5.5
$2 / $10Same ₹204.53 and ₹1,023 on Velona. Cache reads are half the price at $0.10 per 1M. The better default for new projects.
The token price is identical, so moving from Sonnet 5 to Sonnet 5.5 costs nothing extra. If your app reuses a long system prompt or codebase, the cheaper cache reads on Sonnet 5.5 can lower the bill further. Want the full feature comparison with the older generation? See Claude Sonnet 4.5 vs Sonnet 5.
How you pay for Claude from India
Anthropic bills in US dollars and expects an international card. Velona puts a rupee wallet in front of the same models, so the whole flow stays in rupees.
Billing flow
From a UPI payment to a Claude response
Every model on Velona is priced with one published formula:
To be straight about it: if you already have an international card that works with Anthropic, paying direct in dollars skips the 7% platform fee. Velona is built for developers who do not have that card, hit payment declines, or want to start with ₹10 instead of a dollar minimum and use Claude alongside 300+ other models on one key.
Cheaper alternatives when Sonnet is more than you need
For a coding agent, the cost per step adds up fast. These models cost a fraction of Sonnet 5.5 per step and handle many routine tasks well. Try them on your own prompts before switching.
Coding agent step
Cost of one agent step (30,000 tokens in, 1,000 out)
Velona rupee prices on 16 October 2026
DeepSeek V4 Flash · 336 steps per ₹100
Claude Haiku 5.5 · 279 steps per ₹100
GPT-5 Mini · 103 steps per ₹100
Qwen3 Coder · 98 steps per ₹100
Claude Sonnet 5.5 · 14 steps per ₹100
See every model ranked by rupee price on the cheapest AI models page, and quality rankings on the coding leaderboard. For the full DeepSeek breakdown, read DeepSeek price in India.
Call Claude Sonnet from Python in 2 minutes
Velona is OpenAI compatible. Install the OpenAI SDK, point it at Velona, and pick a Claude model ID from the table above.
pip install openai
import os
from openai import OpenAI
client = OpenAI(
api_key=os.environ["VELONA_API_KEY"],
base_url="https://velona.in/v1",
)
response = client.chat.completions.create(
model="anthropic/claude-sonnet-5.5",
messages=[{"role": "user", "content": "Explain Python decorators with one example."}],
max_tokens=500,
)
print(response.choices[0].message.content)
print(response.usage) # tokens billed for this request
Setting max_tokens caps the expensive part of the request. Every call then shows up in your usage dashboard with its exact rupee cost. A ₹100 top up covers about 140 Sonnet 5.5 chat replies.
Need the full India setup, with key creation and curl examples? Read Claude API in India: price in ₹ and how to pay with UPI.
Frequently asked questions
How much does Claude Sonnet 5 cost?
Claude Sonnet 5 costs $2 per 1M input tokens and $10 per 1M output tokens. On Velona that is ₹204.53 and ₹1,023 per 1M tokens, paid in rupees.
Is Claude Sonnet 5 API pricing per token?
Yes. You pay for the tokens you send and the tokens the model writes, priced per million. A typical chat reply of 1,000 input and 500 output tokens costs about ₹0.72 on Sonnet 5 or 5.5.
Is Claude Sonnet 5.5 more expensive than Sonnet 5?
No. Both cost $2 per 1M input tokens and $10 per 1M output tokens. Sonnet 5.5 has cheaper cache reads, at $0.10 per 1M instead of $0.20.
What is the Claude Fable price in India?
Claude Fable 5.1 costs $10 per 1M input tokens and $50 per 1M output tokens. On Velona that is ₹1,023 and ₹5,113 per 1M tokens.
What is the Claude Opus 5.5 price?
Claude Opus 5.5 costs $4 per 1M input tokens and $20 per 1M output tokens, or ₹409.06 and ₹2,045 on Velona.
What is the cheapest Claude model?
Claude Haiku 5.5, at $0.10 per 1M input tokens and $0.50 per 1M output tokens for prompts up to 100K tokens. That is ₹10.23 and ₹51.13 on Velona.
Can I pay for the Claude API with UPI?
Yes, through Velona. You top up a rupee wallet with UPI, RuPay, a domestic card or net banking from ₹10, and call any Claude model with one API key. No international card is needed.
Try Claude Sonnet 5.5 in rupees.
Add ₹100 with UPI and send your first request in two minutes.
How we calculated this
Dollar prices are Anthropic's published API list prices as of 14 October 2026. Rupee prices use Velona's formula at a USD to INR rate of 95.10 on 16 October 2026, and match what Velona billed for real Sonnet 5.5 and Haiku 5.5 requests that day. Request sizes are typical examples, not averages. All figures exclude prompt caching and batch discounts. Prices change, so check the Claude Sonnet 5 page for the live rate.