Claude Opus 5 API Pricing
Anthropic · released 2026-07-24
Also searched as: Opus 5 pricing
Claude Opus 5 price
0.500¢ / 1K
2.50¢ / 1K
1.00¢ / 1K
Official list price from Anthropic, as recorded by models.dev
What it actually costs
| Workload | Tokens | Cost |
|---|---|---|
| Short chat reply | 500 in / 300 out | $0.0100 |
| RAG query | 4,000 in / 500 out | $0.0325 |
| Long document summary | 50,000 in / 1,000 out | $0.2750 |
| Code review of a 2k-line file | 30,000 in / 2,000 out | $0.2000 |
| 1,000 chat replies/day for 30 days | 30,000 × (500 in / 300 out) | $300.00 |
Cost = (input tokens ÷ 1,000,000 × $5.00) + (output tokens ÷ 1,000,000 × $25.00).
Benchmarks
| Benchmark | Value | Source | As of |
|---|---|---|---|
| Epoch Capabilities Index (ECI) | 164.4 | epoch | 2026-07-24 |
| Epoch Mock AIME accuracy | 98.9% | epoch | 2026-07-24 |
| Epoch FrontierMath (T1–3) accuracy | 85.6% | epoch | 2026-07-24 |
| Arena text rating | 1492 | arena | 2026-10-28 |
| Arena coding rating (human preference votes, not task accuracy) | 1534 | arena | 2026-10-28 |
| Arena math rating | 1521 | arena | 2026-10-28 |
| Arena WebDev rating | 1691 | arena | 2026-10-28 |
LiveBench
Shown on this model page only — LiveBench data never appears in any leaderboard, comparison table or structured data on this site.
| LiveBench metric | Value | As of |
|---|---|---|
| LiveBench Global Average | 80.1 | 2026-06-25 |
| LiveBench Reasoning | 91.2 | 2026-06-25 |
| LiveBench Coding | 81.4 | 2026-06-25 |
| LiveBench Agentic Coding | 65.2 | 2026-06-25 |
| LiveBench Mathematics | 95.7 | 2026-06-25 |
| LiveBench Data Analysis | 74.6 | 2026-06-25 |
| LiveBench Language | 88.7 | 2026-06-25 |
| LiveBench Instruction Following | 63.8 | 2026-06-25 |
LiveBench leaderboard (livebench.ai, github.com/LiveBench/new-livebench), CC BY-SA 4.0 — livebench.ai · github.com/LiveBench/new-livebench · CC BY-SA 4.0
Similar ECI, lower price
| Model | ECI | Blended $/1M |
|---|---|---|
| Claude Haiku 5.5 | 160.2 | $0.20 |
| GPT-6.1 Terra | 167.8 | $2.00 |
| Claude Sonnet 5.5 | 166.2 | $4.00 |
| GPT-6 Sol | 164.1 | $4.00 |
| GPT-6.1 Sol | 168.9 | $4.00 |
More from Anthropic
| Model | Released | Blended $/1M |
|---|---|---|
| Claude Haiku 5.5 | 2026-10-07 | $0.20 |
| Claude Sonnet 5.5 | 2026-09-28 | $4.00 |
| Claude Opus 5.5 | 2026-09-22 | $8.00 |
| Claude Mythos 5.1 | 2026-09-01 | $20.00 |
| Claude Fable 5.1 | 2026-09-01 | $20.00 |
| Claude Sonnet 5 | 2026-06-30 | $4.00 |
| Claude Mythos 5 | 2026-06-09 | $20.00 |
| Claude Fable 5 | 2026-06-09 | $20.00 |
Count tokens & estimate cost for Claude Opus 5 See Claude Opus 5 on the leaderboard
Prices: Anthropic's official list price as recorded by models.dev; benchmarks: see sources below; last synced 2026-11-06.
Compare Claude Opus 5 with …
FAQ
Who develops Claude Opus 5?
Claude Opus 5 is developed by Anthropic, released 2026-07-24.
How much does Claude Opus 5 cost per million tokens?
Claude Opus 5 costs $5.00 per 1M input tokens and $25.00 per 1M output tokens ($10.00 blended), per Anthropic's official list price as recorded by models.dev.
What is the Claude Opus 5 API price?
The Claude Opus 5 API price is $5.00 per 1M input tokens and $25.00 per 1M output tokens ($10.00 per 1M blended).
How much does Claude Opus 5 cost for a typical request?
A short chat reply (500 in / 300 out tokens) costs about $0.0100 on Claude Opus 5, at $5.00/1M input and $25.00/1M output.
Is Claude Opus 5 free?
Claude Opus 5's official API list price is $5.00 per 1M input tokens and $25.00 per 1M output tokens. This site tracks list prices only, not free tiers or trial credits — check Anthropic's site for those.
How does Claude Opus 5 rank on benchmarks?
Claude Opus 5 scores 164.4 on the Epoch Capabilities Index (ECI), ranking #8 of 107 models we track with a published ECI score.
Is there a cheaper model with a similar ECI to Claude Opus 5?
Yes — Claude Haiku 5.5 scores a similar ECI (160.2) at $0.20/1M blended, versus $10.00/1M for Claude Opus 5.
Where does this data come from?
Pricing for Claude Opus 5 comes from models.dev and LiteLLM (open MIT-licensed catalogs); benchmark scores come from Epoch AI and LMArena (open CC BY 4.0 licenses); LiveBench scores (below) are CC BY-SA 4.0 — see /data-sources/. Last synced 2026-11-06.