Claude API pricing: every model, every rate
Anthropic prices Claude by capability tier rather than by size. Below is the full rate card as it actually bills, including the legacy generations still running in production.
claude-opus-4-8. Rates checked 2026-11-02.Full rate card
41 billable models in total, 20 of them text models. On those text models: 13 of 20 carry a rate verified against Anthropic's own published pricing (platform.claude.com), the remaining 7 are gateway rates only. USD per 1M tokens, checked 2026-11-02.
| Model | Input | Output | Out/In | Official (in / out) |
|---|---|---|---|---|
claude-3-opus-20240229 | $7.5 | — | $15 / $75 -50.0% | |
claude-4-opus | $7.5 | $37.5 | 5.0× | not verified |
claude-opus-4-1-20250805 | $7.5 | $37.5 | 5.0× | $15 / $75 -50.0% |
claude-opus-4-1-20250805-thinking | $7.5 | $37.5 | 5.0× | not verified |
claude-opus-4-20250514 | $7.5 | $37.5 | 5.0× | $15 / $75 -50.0% |
claude-opus-4-20250514-thinking | $7.5 | $37.5 | 5.0× | not verified |
claude-fable-5 | $5 | $25 | 5.0× | $10 / $50 -50.0% |
claude-fable-5-1 | $5 | $25 | 5.0× | $10 / $50 -50.0% |
claude-2.0 | $4 | — | not verified | |
claude-2.1 | $4 | — | not verified | |
claude-opus-4-5-20251101 | $2.5 | $12.5 | 5.0× | $5 / $25 -50.0% |
claude-opus-4-6 | $2.5 | $12.5 | 5.0× | $5 / $25 -50.0% |
claude-opus-4-6-thinking | $2.5 | $12.5 | 5.0× | not verified |
claude-opus-4-7 | $2.5 | $12.5 | 5.0× | $5 / $25 -50.0% |
claude-opus-4-7-thinking | $2.5 | $12.5 | 5.0× | not verified |
claude-opus-4-8 | $2.5 | $12.5 | 5.0× | $5 / $25 -50.0% |
claude-opus-5 | $2.5 | $12.5 | 5.0× | $5 / $25 -50.0% |
global.anthropic.claude-opus-4-6-v1 | $2.5 | $12.5 | 5.0× | not verified |
claude-opus-5-5 | $2 | $10 | 5.0× | not verified |
anthropic/claude-3.7-sonnet | $1.5 | $7.5 | 5.0× | not verified |
anthropic/claude-sonnet-4 | $1.5 | $7.5 | 5.0× | not verified |
claude-3-5-sonnet-20240620 | $1.5 | $7.5 | 5.0× | not verified |
claude-3-7-sonnet | $1.5 | $7.5 | 5.0× | $3 / $15 -50.0% |
claude-3-7-sonnet-20250219-thinking | $1.5 | $7.5 | 5.0× | not verified |
claude-3-7-sonnet-thinking | $1.5 | $7.5 | 5.0× | not verified |
claude-3-sonnet-20240229 | $1.5 | — | $3 / $15 -50.0% | |
claude-3.7-sonnet-thinking | $1.5 | $7.5 | 5.0× | not verified |
claude-4-sonnet-20250514 | $1.5 | $7.5 | 5.0× | not verified |
claude-sonnet-4-20250514-thinking | $1.5 | $7.5 | 5.0× | not verified |
claude-sonnet-4-5-20250929 | $1.5 | $7.5 | 5.0× | $3 / $15 -50.0% |
claude-sonnet-4-5-20250929-thinking | $1.5 | $7.5 | 5.0× | not verified |
claude-sonnet-4-6 | $1.5 | $7.5 | 5.0× | $3 / $15 -50.0% |
claude-sonnet-5 | $1 | $5 | 5.0× | $2 / $10 -50.0% |
claude-sonnet-5-5 | $1 | $5 | 5.0× | not verified |
claude-instant-1 | $0.815 | — | not verified | |
claude-haiku-4-5-20251001 | $0.5 | $2.5 | 5.0× | $1 / $5 -50.0% |
claude-haiku-4-5-20251001-thinking | $0.5 | $2.5 | 5.0× | not verified |
claude-3-5-haiku-20241022 | $0.4 | $2 | 5.0× | $0.8 / $4 -50.0% |
claude-1-100k | $0.25 | — | not verified | |
claude-3-haiku-20240307 | $0.125 | — | $0.25 / $1.25 -50.0% | |
Claude-haiku-5-5 | $0.05 | $0.25 | 5.0× | not verified |
On Anthropic the Out/In column sits at a flat 5.0× on every model here, it outweighs the input price as soon as output dominates your bill.
Across 20 priced text models the blended rate spans $0.3 (Claude-haiku-5-5) to $45 (claude-opus-4-20250514) per 1M tokens (a 150× spread). At 150× between its own cheapest and dearest, picking the wrong tier inside Anthropic costs more than picking the wrong vendor.
How these rates were verified
18 of the 41 Anthropic models above carry a list price we could confirm on Anthropic’s own documentation; the other 23 are gateway rates only, and no discount is claimed for those. For the confirmed ones the gateway rate currently sits 50% below the published list:
| Model | Official list (in / out) | Here (in / out) | Gap |
|---|---|---|---|
claude-3-5-haiku-20241022 | $0.8 / $4 | $0.4 / $2 | 50% lower |
claude-3-7-sonnet | $3 / $15 | $1.5 / $7.5 | 50% lower |
claude-3-haiku-20240307 | $0.25 / $1.25 | $0.125 / — | 50% lower (input) |
claude-3-opus-20240229 | $15 / $75 | $7.5 / — | 50% lower (input) |
claude-3-sonnet-20240229 | $3 / $15 | $1.5 / — | 50% lower (input) |
claude-fable-5 | $10 / $50 | $5 / $25 | 50% lower |
claude-fable-5-1 | $10 / $50 | $5 / $25 | 50% lower |
claude-haiku-4-5-20251001 | $1 / $5 | $0.5 / $2.5 | 50% lower |
claude-opus-4-1-20250805 | $15 / $75 | $7.5 / $37.5 | 50% lower |
claude-opus-4-20250514 | $15 / $75 | $7.5 / $37.5 | 50% lower |
claude-opus-4-5-20251101 | $5 / $25 | $2.5 / $12.5 | 50% lower |
claude-opus-4-6 | $5 / $25 | $2.5 / $12.5 | 50% lower |
claude-opus-4-7 | $5 / $25 | $2.5 / $12.5 | 50% lower |
claude-opus-4-8 | $5 / $25 | $2.5 / $12.5 | 50% lower |
claude-opus-5 | $5 / $25 | $2.5 / $12.5 | 50% lower |
claude-sonnet-4-5-20250929 | $3 / $15 | $1.5 / $7.5 | 50% lower |
claude-sonnet-4-6 | $3 / $15 | $1.5 / $7.5 | 50% lower |
claude-sonnet-5 | $2 / $10 | $1 / $5 | 50% lower |
List prices confirmed 2026-11-02 against platform.claude.com. Gap is computed on the blended input+output sum, except where the endpoint publishes no output rate for a model — those rows are compared on the input rate and marked accordingly. Gateway pricing follows upstream promotions and can move without notice — confirm on the vendor page before you commit a workload.
What Anthropic costs at three usage levels
Priced on claude-sonnet-5 at $1 in / $5 out; claude-opus-4-8 at $2.5 in / $12.5 out per 1M tokens, with a 5:1 input-to-output ratio, roughly what an
interactive workload produces.
| Usage level | Tokens per month (in / out) | Lowest-cost tierclaude-sonnet-5 | Median-rate modelclaude-opus-4-8 |
|---|---|---|---|
| Light pilot or side project | 5M / 1M | $10.00 | $25.00 |
| Working one product in production | 50M / 10M | $100.00 | $250.00 |
| Heavy high-volume pipeline | 500M / 100M | $1,000.00 | $2,500.00 |
The first column is this vendor's lowest-cost tier; the second is the model closest to its median rate, which is nearer what a real deployment bills. Swap in your own token split in the calculator.
Same budget, other vendors
3 models from 3 other vendors priced within
35% of claude-opus-4-8 on a blended rate (the cheapest, the dearest and one
in between). This is the horizontal check a Anthropic rate page on its own cannot give you.
| Model | Vendor | In / out per 1M | vs this |
|---|---|---|---|
Kimi-K2-Instruct | Moonshot | $2 / $8 | −33.3% |
qwen3-coder | Alibaba | $3 / $12 | same |
deepseek-v4-pro-202606 | DeepSeek | $4.5 / $13.5 | +20.0% |
Blended comparison (input + output per 1M tokens). Price is one axis, none of these is a Anthropic model, so treat them as candidates to test side by side at this budget rather than drop-in substitutes.
Why the output rate decides the bill
Across the 20 priced Anthropic text models, output bills at the same 5.0× the input rate on every one of them. On Anthropic that multiple lands on the smaller half of the token count and the larger half of the bill, because output is what a chat or agent turn actually produces.
Worked out: at a 5.0× multiple, a job whose token count is 80% input still spends 56% of its cost on the output it generates. Setting Anthropic beside any other vendor on the input rate alone hides that.
What Anthropic is good at
Claude is the default choice for agentic coding, long-context document work, and anything where instruction-following quality matters more than raw token cost.
Detailed pages for Anthropic models
The models we cover in depth, each with worked cost at three usage levels and a comparison against similarly priced alternatives.
- Claude Haiku 5.5 — $0.05 / $0.25 per 1M tokens
- Claude Haiku 4.5 — $0.5 / $2.5 per 1M tokens
- Claude Sonnet 5 — $1 / $5 per 1M tokens
- Claude Sonnet 5.5 — $1 / $5 per 1M tokens
- Claude Sonnet 4.6 — $1.5 / $7.5 per 1M tokens
- Claude Opus 5.5 — $2 / $10 per 1M tokens
- Claude Opus 5 — $2.5 / $12.5 per 1M tokens
- Claude Opus 4.8 — $2.5 / $12.5 per 1M tokens
- Claude Fable 5 — $5 / $25 per 1M tokens
Routing these rates into your tools
Every model above is reachable from the same OpenAI-compatible endpoint, so the rates on this page apply wherever you already work. These are the tools we have profiled end to end — Anthropic models included:
| Tool | What to set |
|---|---|
| LangChain | one base_url kwarg — then check whether token usage still comes back |
| LlamaIndex | OpenAILike for gateway IDs, and a second object for the embedding model |
| AutoGen | declare function_calling or tool use fails silently |
| OpenAI Agents SDK | tracing exports to OpenAI by default; the default transport is Responses |
| Codex CLI | 0.122+ rejects wire_api = "chat" at config load |
| Gemini CLI | a model without tool calls replies in prose and never touches the repo |
| CrewAI | custom_openai=True, and cost is agents × turns |
| DSPy | the bill is in compile(), not in inference |
| Cursor | override the base URL and bill your own key |
| LiteLLM model names | the slash is a routing instruction — openai/, or nothing is sent |
| streaming calls | stream_options, or every token count comes back null |
| an empty model picker | read /v1/models and type the id by hand |
| long sessions | two caps, not one — the window and the maximum output |
| every tool we have profiled | the full matrix of what to set and where |
Every one of these takes the same two values (a base URL and a key) and none of them require changing your code beyond where the client is constructed. Pointed at Anthropic, that is the whole change. The differences that matter are in each tool's defaults, which is what each guide covers.
Other vendors
Every rate card here is built the same way, from the same price snapshot (2026-11-02), so the numbers are comparable across vendors:
OpenAI · Google · DeepSeek · Alibaba · xAI · Zhipu · Moonshot · Meta · MiniMax · ByteDance
FAQ
How much does Anthropic charge per million tokens?
Across the 20 priced text models in this catalogue, the blended rate (input plus output per 1M tokens) runs from $0.3 on Claude-haiku-5-5 to $45 on claude-opus-4-20250514, as of 2026-11-02. That 150× spread is what matters: inside Anthropic alone, picking the wrong tier moves the bill further than switching to another vendor would.
What is the cheapest Anthropic model for high-volume work?
Claude-haiku-5-5 is the lowest blended rate here at $0.05 input and $0.25 output per 1M tokens. On a Heavy workload (500M input and 100M output tokens a month) that bills about $50.00. At the other end, the same workload on claude-opus-4-20250514 costs about $7,500.00.
Is Anthropic cheaper through a gateway than buying direct?
For the 13 models where a published Anthropic list price could be verified, the gateway rate is below list by between 50% and 50%, checked 2026-11-02. The rate is not fixed, it moves with upstream promotions, and several models in this catalogue have no published list price at all, so no comparison is claimed for them.
How does output pricing change a Anthropic bill?
On Anthropic output is billed at 5.0× to 5.0× the input rate, median 5.0×, so the input rate alone never predicts the bill. At 5.0× even a workload that is 80% input by token count lands most of its cost on the output column. Price your own split against claude-opus-4-8's $2.5 input rate in the calculator rather than extrapolating from that headline.
What is the Anthropic API price per 1M tokens?
Anthropic quotes every model separately for input and output. claude-opus-4-8 (the one closest to this vendor's median rate) is $2.5 per 1M input and $12.5 per 1M output as of 2026-11-02. The lowest tier in this catalogue is Claude-haiku-5-5, at $0.05 in and $0.25 out.
How much does Anthropic cost for a coding agent?
A heavy agent month — 500M input and 100M output tokens, the shape a coding assistant really produces because every turn resends the whole context, bills about $2,500.00 on claude-opus-4-8. Output decides most of that figure: on Anthropic the median model bills output at 5.0× its input rate, so extrapolating from the headline input rate understates the bill by close to that factor.
Is Sonnet 5 or Opus 5 the better default for an agent loop?
Purely on price, Sonnet 5 is the default: it bills at roughly half of Opus 5 on both input and output. Opus 5 only earns its premium on turns where a wrong answer costs more than the tokens do. The common pattern is Opus 5 for planning and Sonnet 5 for the execution steps it emits.
See all 38 models → or use the cost calculator with your own token split.