ARCHIVE PRICEDEPRECATED MODELTEXT + VISIONPROMPT CACHINGBATCH -50%
Claude Opus 4 API Pricing
Archive profile for Claude Opus 4: $15/M input, $75/M output, and $1.5/M cache-hit input. Anthropic marks Opus 4 as deprecated, with a scheduled retirement on June 15, 2026. Pulled directly from platform.claude.com daily.
Archived Input - per 1M tokens
$15.00/M
Published Anthropic table deprecated
Archived Output - per 1M tokens
$75.00/M
Use for invoice checks deprecated
Cached input - 90% off
$1.50/M
Cache 5min or 1h -90%
Effective - agentic blend
$9.62/M
92/8 split - 82% cache
§ 01 / TERMINAL
Run the numbers.
Archive calculator pre-loaded with Claude Opus 4 rates. Tweak spend, output mix, or cache hit rate; share the URL to share the calculation.
$ /mo
Workload split
Prompt cache hit rate
Tokens you can process
—
Words equivalent (English)
—
Effective rate
—
§ 02 / SCENARIOS
Real-world presets.
CODING AGENT ARCHIVE
Repo-wide bug fix
$0.843/task
LONG DOC ANALYSIS ARCHIVE
Reading 100-page contracts
$1.29/doc
RAG SUPPORT ARCHIVE
Support agent ticket triage
$0.053/ticket
ASSISTANT ARCHIVE
Research planning turn
$0.160/turn
§ 03 / TAPE
Price history.
Input · $15/M
Output · $75/M
Cached · $1.5/M
MAY 14 Launch at $15/M - $75/MMAY 18 Deprecated; retirement scheduled for June 15, 2026
§ 04 / TOKENIZER
Paste text. See tokens. See cost.
Estimate · anthropic-bpe-estimate · ≈3.5 chars/token Auto-counts as you type
This is a chars-per-token approximation, not a real tokenizer. Actual tokens vary by language, code density, and tool-call overhead — counts are typically ±10–20% off for English prose, more for code or non-Latin scripts. For exact billing, use the vendor's official tokenizer.
Characters —
Words —
Tokens (estimated) —
Cost as input · uncached —
Cost as output · uncached —
Cost as cached input —
| Model | Input /M | Output /M | Effective blended | Context | Best for |
|---|---|---|---|---|---|
| Claude Opus 4.7 | $5.00 cache $0.50 | $25.00 | $3.21 cheaper | 1M | Frontier reasoning and hard code |
| Claude Opus 4.6 | $5.00 cache $0.50 | $25.00 | $3.21 cheaper | 1M | Frontier reasoning and hard code |
| Claude Opus 4.5 | $5.00 cache $0.50 | $25.00 | $3.21 cheaper | 200K | Frontier reasoning and hard code |
| Claude Opus 4.1 | $15.00 cache $1.50 | $75.00 | $9.62 same blend | 200K | Legacy Opus workloads |
| Claude Sonnet 4.6 | $3.00 cache $0.30 | $15.00 | $1.92 cheaper | 1M | Production agents and coding |
| Claude Sonnet 4.5 | $3.00 cache $0.30 | $15.00 | $1.92 cheaper | 200K | Production agents and coding |
| Claude Haiku 4.5 | $1.00 cache $0.10 | $5.00 | $0.64 cheaper | 200K | Support and classification |
| GPT-5.4 | $2.50 cache $0.25 | $15.00 | $1.80 cheaper | 1.05M | Tool use and app agents |
| Gemini 2.5 Pro | $1.25 cache $0.13 | $10.00 | $1.10 cheaper | 2M | Long-context document analysis |
Frequently asked.
Practical pricing questions, with vendor list prices separated from workload assumptions.
Q · 01 Is Claude Opus 4 still recommended for new API usage? +
Anthropic lists this model as deprecated, with retirement scheduled for June 15, 2026. The current published price is still
$15/M input and $75/M output, but Anthropic recommends migrating to claude-opus-4-7.Q · 02 What should I migrate Claude Opus 4 workloads to? +
Use claude-opus-4-7 for new production work. Anthropic's deprecation table names that replacement, while this archive keeps the old list price for invoice checks and migration comparisons.
Q · 03 How does prompt caching change the price? +
Cache hits are listed at
$1.5/M, which is 10% of the $15/M base input rate. Anthropic also charges cache writes at $18.75/M for 5-minute writes and $30/M for 1-hour writes. The calculator models repeated prompt sections as cache hits.Q · 04 Is there a Batch API discount? +
Yes. Anthropic's Batch API gives a 50% discount on input and output tokens. For Claude Opus 4, the batch table maps to
$7.5/M input and $37.5/M output.Q · 05 Does long context cost extra? +
This page uses the model's documented
200K context label. Longer-context Opus 4.6, Opus 4.7, and Sonnet 4.6 use standard pricing across the full 1M window. Prompt caching and batch discounts still apply according to their normal rules.Q · 06 Does regional pricing differ? +
Anthropic says earlier models do not support the
inference_geo parameter and use standard first-party API pricing. Bedrock and Vertex AI publish their own regional pricing policies.Q · 07 Are volume discounts published? +
No fixed public volume-discount ladder is published. Enterprise and committed-spend contracts can have private terms, but standard API users should assume the list prices shown here unless they have a separate agreement.
Q · 08 How accurate is the tokenizer estimate? +
The live counter uses a Claude-family English estimate of
4.875 characters per token. Actual billing uses Anthropic's server-side tokenizer; code, tables, and non-English text can differ materially.