by Anthropic

Claude Opus 4.7.

text closed 200K ctx Transformer Quality 83.1 Elo 1387

Cheapest input

$5.0/M

on DeepInfra

Cheapest output

$25.0/M

on DeepInfra

Fastest

37 tok/s

on OpenRouter

Hosted equiv.

~$9.0/hr

@ 100 tok/s on DeepInfra

Frontier reasoning and long-form coding from Anthropic.

Hosted API only

No self-host path — closed weights.

Claude Opus 4.7's weights aren't published. Use it via the access providers below.

Where to use it Top 5 cheapest · 6 total below

Cheapest hosted endpoints.

Provider	Access	$/M in	$/M out
DeepInfra	hosted inference	$5.0	$25.0	Launch ↗
OpenRouter	api aggregator	$5.0	$25.0	Launch ↗
Anthropic API	api direct	$15.0	$75.0	Launch ↗
AWS Bedrock	api aggregator	$15.0	$75.0
Google Cloud Vertex AI	api aggregator	$15.0	$75.0

Capability snapshot Full benchmarks →

What it's best at.

Coding 92.0

General knowledge 88.0

Math 87.5

Reasoning 84.0

Performance

Speed across providers.

Tokens/sec and time-to-first-token measured against the same prompt template on each provider's API.

Provider	Tokens/sec	TTFT	Total
OpenRouter	37.5	1433 ms	3655 ms

Sources

Official references.

Launch post ↗ Demo / Chat ↗ System card ↗ Docs ↗

Family

Variants in the Claude family.

Claude Sonnet 4.6

closed proprietary

Best price-performance from Anthropic. Default for production agents.

Claude 3.5 Sonnet

closed proprietary

Anthropic's 3.5 generation — still in active production.

Claude Haiku 4.5

closed proprietary

Fast, cheap Claude variant for high-throughput inference.

Claude 3.5 Haiku

closed proprietary

Fast/cheap Claude 3.5 variant — production fallback for Haiku 4.5.

Best for

Workloads.

Run LLMs (inference / serving) General ML research and experiments

FAQ

Frequently asked.

How do I run Claude Opus 4.7?

Claude Opus 4.7 is a closed-source API model. The cheapest way to access it is through the API providers listed on this page (direct API, aggregators, and hosted chat UIs).

Where can I access Claude Opus 4.7?

Claude Opus 4.7 is available via Anthropic API, Claude.ai, AWS Bedrock, Google Cloud Vertex AI, DeepInfra. Each access option lists its own pricing (per million tokens or hourly hosting).

How much does it cost to run Claude Opus 4.7?

API pricing starts at $15.0/M input tokens and $75.0/M output tokens. Self-hosting cost depends on the GPU you rent — see the Run It Yourself tab.

Is Claude Opus 4.7 open-source or proprietary?

Claude Opus 4.7 is a proprietary model from Anthropic. Access is API-only — there are no public weights to download.

API pricing

What it costs per month across providers.

Estimate your monthly bill for Claude Opus 4.7 across every host that publishes per-token pricing. Slide your token volumes; the chart + table re-rank cheapest-first.

Cheapest

$100.0

OpenRouter

Most expensive

$300.0

Google Vertex AI

Spread

$200.0

max − min

Providers

with priced rows

Monthly bill

Cheapest provider on the left.

Total monthly cost — input + output tokens combined.

Per provider

Bill breakdown.

Provider	$/M in	$/M out	Input cost	Output cost	Monthly total
OpenRouter	$5.0	$25.0	$50.0	$50.0	$100.0	Sign up ↗
DeepInfra	$5.0	$25.0	$50.0	$50.0	$100.0	Sign up ↗
Anthropic API	$15.0	$75.0	$150.0	$150.0	$300.0	Sign up ↗
AWS Bedrock	$15.0	$75.0	$150.0	$150.0	$300.0
Google Vertex AI	$15.0	$75.0	$150.0	$150.0	$300.0

Full calculator

Want to compare token volumes across other models too? Open the standalone API pricing tool →

Capability snapshot

What it's best at.

Coding 92.0

General knowledge 88.0

Math 87.5

Reasoning 84.0

Scores normalised against benchmark ceilings (100 = perfect). Coloured by tier — coral 80+ frontier, lavender 65+ strong, sage 50+ solid, slate below.

Benchmarks

Published scores.

Benchmark	Score	Source
GPQA	67.0	official ↗
MATH	87.5	official ↗
MMLU	88.0	official ↗
MMLU-Pro	84.0	official ↗
HumanEval	92.0	official ↗
SWE-bench	72.5	official ↗

Leaderboard standing

Independent rankings.

LMSYS Chatbot Arena

1387

Rank #2 · Elo from blind head-to-head votes

View leaderboard ↗

Artificial Analysis Quality Index

86.0

Composite of reasoning + coding + tool-use benchmarks

View on Artificial Analysis ↗

Description

About Claude Opus 4.7.

Claude Opus 4.7 is Anthropic's frontier model for complex reasoning, long-form coding, and agentic tasks. The 4.x series introduced extended thinking, which lets Claude allocate compute proportional to problem difficulty — short questions stay fast, hard ones get deeper deliberation. Opus 4.7 leads its family on coding benchmarks (SWE-bench Verified) and tops the LMSYS Arena Hard leaderboard for technical prompts. It supports 200K-token context, vision inputs, tool use, and JSON mode. Most users access it through the Anthropic API or Claude.ai; Bedrock and Vertex AI mirror the same model behind enterprise contracts. Opus is the slowest and most expensive tier of Claude — for routine workloads, Sonnet typically wins on price-performance.

Architecture