Claude Opus 5: Anthropic's Model for Agentic Coding and Enterprise Work
Key facts · verified Oct 30, 2026
Claude Opus 5 is an Anthropic model released on 24 Jul 2026. Its API price is $5 per million input tokens and $25 per million output tokens. Its context window is 1M tokens. It ranks #14 on LMArena's text leaderboard (Elo 1492, read 1 Nov 2026). Source: platform.claude.com.
Claude Opus 5 is an Anthropic model for complex agentic coding and enterprise work, released on July 24, 2026 with a 1M-token context window, adaptive thinking and 128K maximum output, priced at $5 per million input tokens and $25 per million output tokens. It superseded Claude Opus 4.8. Anthropic lists Claude Opus 5 as a legacy model, still available, since Claude Opus 5.5 was released on September 22, 2026.
Key Features
What is Claude Opus 5?
Claude Opus 5 is the model Anthropic positions for complex agentic coding and enterprise work. In Anthropic's current line-up it sits alongside Claude Fable 5, Claude Sonnet 5 and Claude Haiku 4.5, and it supersedes Claude Opus 4.8, which Anthropic now lists under legacy models. It takes a 1 million token context window and can return up to 128,000 output tokens in a single response.
How Claude Opus 5 Works
Anthropic's model documentation lists Opus 5 as supporting adaptive thinking, and marks the older "extended thinking" mode as not applicable to it. It accepts text and image input and supports tool use. Anthropic does not publish the architecture, parameter count or training details, so this guide does not describe them.
Benchmark Performance
On the LMArena text leaderboard, a head-to-head human-preference ranking, Claude Opus 5 has 1492 Elo (rank 14, high effort); Anthropic's Claude Fable 5 has 1504 Elo (rank 6, high effort). Arena measures blind human preference across open-ended prompts; it is not a coding benchmark. This guide does not quote MMLU, SWE-bench or GPQA figures because each lab runs those under its own harness.
Pricing and Access
Through the Anthropic API, Claude Opus 5 is priced at $5 per million input tokens and $25 per million output tokens. On the consumer side, Anthropic's published plans are: Free at $0; Pro at $20/month billed monthly, or $17/month if billed annually ($200 up front); Max from $100/month; Team at $25 per seat/month monthly, or $20 billed annually; and Enterprise at a seat price plus usage at API rates. Anthropic's pricing page does not state which models each consumer tier unlocks, so no mapping is given here.
Weights
Anthropic does not publish weights for Claude Opus 5.
Ranking and Price Against Other Models
On LMArena's text leaderboard Claude Opus 5 has 1492 Elo (rank 14, high effort), Gemini 3.1 Pro 1487 Elo (rank 20) and GPT-5.6 Sol 1485 Elo (rank 22, xhigh effort). Published API prices per million input / output tokens: Claude Opus 5 $5 / $25, Gemini 3.1 Pro $2 / $12, GPT-5.6 Sol $4 / $20 (OpenAI's promotional price).
Key Takeaways
- Anthropic positions Opus 5 for complex agentic coding and enterprise work
- 1M token context window, 128k maximum output tokens
- API pricing $5 / $25 per million input / output tokens
- Claude Pro is $20/month billed monthly, or $17/month billed annually
- LMArena: 1492 Elo (rank 14, high effort)
- Supersedes Claude Opus 4.8, which Anthropic now lists as a legacy model
Official Resources
Full Specifications
$5 / $25 per million tokens | |
|---|---|
| Identity | |
| Developer | Anthropic |
| Released | Jul 2026 |
| Status | Available (legacy) |
| Licence | Proprietary |
| Self-hostable | No |
| Cost | |
| Blended $/1M tokensinput × 0.75 + output × 0.25 | $10.000 / 1M tokens |
| Input price | $5.000 / 1M tokens |
| Output price | $25.000 / 1M tokens |
| Cached input | $0.50 / 1M tokens (cache hit) |
| Batch discount | 50% off input/output (Batch API) |
| Free tier | Small trial credit for new accounts, no ongoing free tier |
| Capacity | |
| Context window | 1M tokens |
| Max output | 128k tokens |
| Long-context surcharge | None (full 1M window at standard pricing) |
| Capability | |
| Vision in | Yes |
| Audio in | No |
| Function calling | Yes |
| Structured output | Yes |
| Extended reasoning | Yes |
| Web search | Yes |
| Code execution | Yes |
| Access | |
| API | Yes |
| Chat app | Yes |
| Cloud marketplaces | AWS Bedrock, Google Cloud, Microsoft Foundry |
| Fine-tuning | No |
| Measured quality | |
| LMArena Elo | 1492 (checked Nov 2026) |
| LMArena rank | Rank 14 |
| GPQA Diamond | 93 |
| Elo per dollarLMArena Elo ÷ blended $/1M tokens | 149 |