Short answer
Claude API pricing as of October 26, 2026, per million tokens (input / output): Claude Opus 5.5 costs $4 / $20, Sonnet 5.5 $2 / $10, Haiku 5.5 $0.10 / $0.50 for prompts up to 100K tokens, and Fable 5.1 $10 / $50. Cache reads cost a small fraction of input ($0.20 on Opus 5.5), and the Batch API takes 50% off input and output. A typical Claude Code task on Opus 5.5 comes to about $0.50 at these prices. On TeamoRouter the same models cost about 30–35% of Anthropic's list price (live prices are on the pricing page), billed per token with no subscription.
No card at all? You can also top up with USDT, USDC or ETH: click Pay with crypto in the top-up dialog and send it from Binance, OKX or any wallet. The minimum is $5 and no browser extension is needed. Step by step: Pay for Claude and GPT APIs with crypto (USDT).
Claude API prices per million tokens
Anthropic list prices in USD per million tokens, standard mode, from Anthropic's pricing page. Cache write is the 5-minute cache.
| Model | Model ID | Input | Output | Cache read | Cache write (5 min) | Context window |
|---|---|---|---|---|---|---|
| Claude Fable 5.1 | claude-fable-5-1 |
$10 | $50 | $0.25 | $12.50 | 1M |
| Claude Opus 5.5 | claude-opus-5-5 |
$4 | $20 | $0.20 | $5 | 1M |
| Claude Sonnet 5.5 | claude-sonnet-5-5 |
$2 | $10 | $0.10 | $2.50 | 1M |
| Claude Haiku 5.5, prompts ≤ 100K | claude-haiku-5-5 |
$0.10 | $0.50 | $0.01 | $0.125 | 1M |
| Claude Haiku 5.5, prompts > 100K | claude-haiku-5-5 |
$0.50 | $2.50 | $0.05 | $0.625 | 1M |
Anthropic suggests starting with Opus 5.5 for most workloads and moving to Fable 5.1 for demanding reasoning and long-horizon agentic work. Haiku 5.5 is meant for high-volume, latency-sensitive jobs such as classification, extraction and routing. Claude Fable 5.5 is not released as of October 26, 2026 (release status).
Older models still available
| Model | Model ID | Input | Output | Cache read | Cache write (5 min) | Context window |
|---|---|---|---|---|---|---|
| Claude Fable 5 | claude-fable-5 |
$10 | $50 | $1 | $12.50 | 1M |
| Claude Opus 5 | claude-opus-5 |
$5 | $25 | $0.50 | $6.25 | 1M |
| Claude Sonnet 5 | claude-sonnet-5 |
$2 | $10 | $0.20 | $2.50 | 1M |
| Claude Opus 4.8 / 4.7 / 4.6 | claude-opus-4-8, claude-opus-4-7, claude-opus-4-6 |
$5 | $25 | $0.50 | $6.25 | 1M |
| Claude Sonnet 4.6 | claude-sonnet-4-6 |
$3 | $15 | $0.30 | $3.75 | 1M |
| Claude Haiku 4.5 | claude-haiku-4-5 |
$1 | $5 | $0.10 | $1.25 | 200K |
Older does not mean cheaper. Opus 5 and the Opus 4.x models cost more per token than Opus 5.5, and Fable 5 cache reads cost four times as much as on Fable 5.1. Sonnet 5's $2 / $10 started as an introductory price and is now its standard price.
When you compare generations, also count tokens, not just prices: Claude 4.7 and later models use a newer tokenizer that produces roughly 30% more tokens for the same text than Sonnet 4.6, Opus 4.6 and earlier.
How caching, batch and long context change the price
Prompt caching. Cache prices are multipliers of the model's base input price:
| Operation | Multiplier | Opus 5.5 |
|---|---|---|
| 5-minute cache write | 1.25× input | $5 |
| 1-hour cache write | 2× input | $8 |
| Cache read | 0.1× input; 0.05× on Opus 5.5 and Sonnet 5.5; 0.025× on Fable 5.1 | $0.20 |
A 5-minute write pays for itself after one cache read, a 1-hour write after two. The 1-hour write costs $20 on Fable 5.1, $8 on Opus 5.5, $4 on Sonnet 5.5 and $0.20 on Haiku 5.5 (prompts up to 100K).
Batch API. Asynchronous batch requests are 50% off input and output: Fable 5.1 $5 / $25, Opus 5.5 $2 / $10, Sonnet 5.5 $1 / $5, Haiku 5.5 $0.05 / $0.25 (prompts up to 100K). Caching multipliers stack with the batch discount.
Long context. Claude 4.6 and later models, except Haiku 5.5, bill the full 1M-token window at standard rates: a 900K-token request costs the same per token as a 9K-token one. On Haiku 5.5, a request whose prompt is over 100,000 tokens, cache reads and writes included, pays five times the base rates ($0.50 / $2.50). Each request is priced on its own.
Fast mode. Opus 5.5 fast mode costs 2× standard, $8 / $40, for up to 2.5× faster output. Caching multipliers apply on top of it, and it can't be combined with the Batch API.
US-only inference. Setting inference_geo to "us" on Claude 4.6 and later models multiplies every token price by 1.1.
What real usage costs
Take one realistic Claude Code task: about 15 turns, 300K input tokens in total, of which 270K are cache hits and 30K are new context written to the cache, plus 15K output tokens. Assuming every request stays under 100K tokens, Haiku 5.5 is billed at its lower rate.
| Fable 5.1 | Opus 5.5 | Sonnet 5.5 | Haiku 5.5 | |
|---|---|---|---|---|
| Cache reads (270K) | $0.068 | $0.054 | $0.027 | $0.0027 |
| Cache writes (30K) | $0.375 | $0.15 | $0.075 | $0.0038 |
| Output (15K) | $0.75 | $0.30 | $0.15 | $0.0075 |
| Per task, list price | ≈ $1.19 | ≈ $0.50 | ≈ $0.25 | ≈ $0.014 |
| Per task, TeamoRouter | ≈ $0.36–0.42 | ≈ $0.15–0.18 | ≈ $0.08–0.09 | ≈ $0.004–0.005 |
Without caching, the same Opus 5.5 task costs about $1.50, three times as much. Caching cuts this task's cost by two-thirds; moving from Opus 5.5 to Sonnet 5.5 cuts it in half.
Scaled to a month:
| Usage | Tasks/month | Opus 5.5, list | Opus 5.5, TeamoRouter | Sonnet 5.5, list | Sonnet 5.5, TeamoRouter |
|---|---|---|---|---|---|
| Light: 3 tasks a day, 20 days | 60 | ≈ $30 | ≈ $9–11 | ≈ $15 | ≈ $4.5–5.3 |
| Daily: 20 tasks a day, 22 days | 440 | ≈ $222 | ≈ $67–78 | ≈ $111 | ≈ $33–39 |
| Heavy: 60 tasks a day, 22 days | 1,320 | ≈ $665 | ≈ $200–233 | ≈ $333 | ≈ $100–117 |
For a reference point, Anthropic reports that across enterprise Claude Code deployments the average is around $13 per developer per active day and $150–250 per developer per month (Claude Code docs). In a running session, /usage shows the cost so far, computed at list price.
Claude API pricing on TeamoRouter
TeamoRouter serves the same Claude models through Anthropic's native Messages API at https://api.teamorouter.com, with one key for every model. Our prices move during the day, so the table shows them relative to Anthropic's list price (per million tokens):
| Model | List price, input / output | TeamoRouter |
|---|---|---|
| Claude Fable 5.1 | $10 / $50 | ≈30–35% of list |
| Claude Opus 5.5 | $4 / $20 | ≈30–35% of list |
| Claude Sonnet 5.5 | $2 / $10 | ≈30–35% of list |
| Claude Haiku 5.5, prompts ≤ 100K | $0.10 / $0.50 | ≈30–35% of list |
The same share applies to cache reads and cache writes. Older models (Fable 5, Opus 5 and 4.x, Sonnet 5, Sonnet 4.6, Haiku 4.5) are also sold below list price. Live prices for every model are on the pricing page.
Billing uses the same token categories as Anthropic (input, output, cache read, cache write), and Haiku 5.5 input and output cost 5× for prompts over 100K tokens here too. Cache hits are billed at the cache-read rate, which is why the Opus 5.5 task above comes to about $0.15–0.18. Claude Code's /usage still shows the list-price figure; your TeamoRouter balance shows what was actually charged.
You top up a balance from $5 by card or Alipay and pay only for the tokens you use. There is no subscription. TeamoRouter offers 50% off on your first $25 of usage. You can also click Pay with crypto in the top-up dialog and pay with USDT, USDC or ETH from Binance, OKX or any exchange or wallet, with no browser extension. BEP20 works from $5, ERC20 and Base from $10, TRC20 from $30, and each invoice is valid for 60 minutes (how to pay with USDT).
To run Claude Code on these prices, create a key in the console and set:
export ANTHROPIC_BASE_URL=https://api.teamorouter.com
export ANTHROPIC_AUTH_TOKEN="your TeamoRouter key"
export ANTHROPIC_API_KEY=""
export ANTHROPIC_MODEL=claude-sonnet-5-5
export ANTHROPIC_DEFAULT_OPUS_MODEL=claude-opus-5-5
export ANTHROPIC_DEFAULT_HAIKU_MODEL=claude-haiku-5-5
claude
Don't add /v1 to the base URL; Claude Code adds the path itself. ANTHROPIC_API_KEY stays empty so an Anthropic key from an earlier setup isn't used by mistake. This setup makes Sonnet 5.5 the default, /model switches to Opus 5.5 for hard tasks, and Claude Code's background chores run on Haiku 5.5. If Claude Code isn't installed yet, follow the installation guide.
FAQ
How much does the Claude API cost per million tokens? At Anthropic's list price: Opus 5.5 $4 input / $20 output, Sonnet 5.5 $2 / $10, Haiku 5.5 $0.10 / $0.50 for prompts up to 100K tokens, and Fable 5.1 $10 / $50. Cache reads, batch requests and fast mode change these rates as described above. On TeamoRouter, the same models cost about 30–35% of list price; live prices are on the pricing page.
What is the cheapest Claude model? Claude Haiku 5.5: $0.10 / $0.50 for prompts up to 100K tokens and $0.50 / $2.50 above that. Both tiers cost less than Haiku 4.5 ($1 / $5), and Haiku 5.5 has a 1M-token context window against Haiku 4.5's 200K. Details in the Haiku 5.5 guide.
Is the Claude API cheaper than Claude Pro or Max? It depends on volume. Pro is $20 a month and Max $100 or $200, and both include Claude Code within usage limits. At list price, 440 Opus 5.5 tasks a month cost about $222, close to Max 20x. The API wins when you hit plan limits, use tools besides Claude Code, or can't subscribe. Full comparison: Claude subscriptions vs API.
Does Claude Code use API pricing? With an API key, yes: every request is billed per token at the rates above. With a Pro or Max login, usage is included in the plan and counts against its limits instead.
Do I need an Anthropic account? For Anthropic's own API, yes: you buy credits in the Claude Console with a credit card. For TeamoRouter, no. You sign up, top up from $5 by card or Alipay, and use one key for every Claude model. Crypto top-ups work too, from $5 on BEP20.
Which model should I use in Claude Code to keep costs down? Keep Sonnet 5.5 as the default at half the Opus 5.5 price, switch to Opus 5.5 with /model for hard tasks, and point the haiku slot at Haiku 5.5 for background chores. The environment variables above do exactly that.
Next steps
- Claude Opus 5.5 API: model ID, prices and Claude Code setup.
- Claude Sonnet 5.5 API: the everyday default model.
- Claude Haiku 5.5 API: the low-cost model and its 100K pricing tier.
- Claude subscriptions vs API: Pro, Max or pay per token.
- Install Claude Code: macOS, Windows and Linux.
- Pay for Claude with USDT: step-by-step crypto top-up.
- Live prices and create an API key.