News · Anthropic · September 22, 2026
Published . By llmwise, AI-assisted.
Claude Opus 5.5 is out: what changed, and what it costs you
Anthropic released Claude Opus 5.5 on September 22, 2026, at $4 per million input tokens and $20 per million output tokens, 20% less than Opus 5. Claude Opus 5.5 has been in llmwise since September 25, 2026: up to 62 Claude Opus 5.5 messages a month on Pro ($20).
Facts checked against Anthropic: Introducing Claude Opus 5.5. Updated .
What Anthropic announced
Each point is from Anthropic's own announcement unless its source says otherwise; the numbers link to the sources below.
Anthropic calls Opus 5.5 the first model in its new Claude 5.5 family, and says it performs at the level of Claude Fable 5.1 on most work.[1]
On typical workloads at default settings, Anthropic's tests put Opus 5.5's cost 40% below Opus 5's: it costs less per token and uses fewer tokens per task.[1]
Cache reads cost $0.20 per million tokens, 60% less than on Opus 5, and cache writes $5 per million.[1]
A fast mode in Claude Code and on the Claude Platform runs up to 2.5 times faster, at $8 per million input tokens and $40 per million output.[1]
Anthropic says Opus 5.5 writes its output more than 30% faster than Opus 5.[1]
Alongside the launch, Anthropic raised the five-hour usage limits on its Pro, Max, Team and seat-based Enterprise plans.[1]
Claude subscribers also got a rate limit reset that they can save and use whenever they choose.[1]
Unlike earlier Opus models, Opus 5.5 is no longer available with thinking switched off.[1]
It's on the Claude Platform as claude-opus-5-5, and through Amazon Web Services, Google Cloud and Microsoft Azure.[1]
Anthropic says Claude Sonnet 5.5 and Claude Haiku 5.5 will follow in the coming weeks.[1]
Can you use Claude Opus 5.5 in llmwise?
What Claude Opus 5.5 costs you in llmwise, from the plans themselves.
Claude Opus 5.5
Claude Opus 5.5 has been in llmwise since September 25, 2026, in place of Claude Opus 5. llmwise's catalog prices it at $4.00 per million tokens in and $20.00 out (Anthropic's list price). But a Claude Opus 5.5 message isn't billed by the token: the table below has its count on every plan.
- Claude Opus 5.5: every limit, plan by plan
- Claude Fable 5.1 vs Claude Opus 5.5, side by side
- Claude Opus 5.5 vs Claude Sonnet 5, side by side
- Claude Opus 5.5 vs Claude Sonnet 5.5, side by side
- Gemini 4 Argon vs Claude Opus 5.5, side by side
- Claude Opus 5.5 vs GPT-6 Astra, side by side
- Claude Opus 5.5 vs GPT-6 Sol, side by side
- Claude Opus 5.5 vs DeepSeek V4 Pro, side by side
- Claude Opus 5.5 vs Claude Haiku 4.5, side by side
- Claude Opus 5.5 vs Gemini 3.1 Pro (preview), side by side
- Claude Opus 5.5 vs Kimi K3, side by side
- Claude API pricing
| Plan | Claude Opus 5.5 |
|---|---|
| Free ($0) | Not in the trial |
| Plus ($10) | Up to 31 a month, about $0.32 each |
| Pro ($20) | Up to 62 a month, about $0.32 each |
| Max ($50) | Up to 200 a month, about $0.25 each |
| Ultra ($100) | Up to 450 a month, about $0.22 each |
| Studio ($200) | Up to 1,000 a month, about $0.20 each |
A paid plan's month is one allowance the models share. So a monthly count is the most you get if the month goes to that model, and “each” is the plan's price divided by it. Every message's price is shown before you send it. Every plan.
The counts above come with each plan's fair-use limit, in AI cost a month: Plus's is $3.75, Pro's $7.50, Max's $20, Ultra's $42, Studio's $85. That leaves room for every message at typical sizes, less if messages run very long or research-heavy. Every limit, explained.
How did Claude Opus 5.5 do on our test prompts?
Beside Gemini 4 Argon and Kimi K3, the models from other makers nearest in price per message.
| Prompt kind | Claude Opus 5.5 | Gemini 4 Argon | Kimi K3 |
|---|---|---|---|
| Writing | 4 of 5 | 4 of 5 | 4 of 5 |
| Summarization | 5 of 5 | 5 of 5 | 3 of 5 |
| Support replies | 5 of 5 | 3 of 5 | 4 of 5 |
| In total | 49 of 50 | 47 of 50 | 46 of 50 |
All passed every prompt on coding, math, data analysis, translation, SQL, answering from documents, and tool use. Through OpenRouter, in our test runs of September 27, 28, 29 and October 2, 7, 8, 9, 15, 2026; every model got the same prompts. Read the replies.
Sources
- [1] Anthropic: Introducing Claude Opus 5.5, published .
Also from primary sources: Anthropic's API pricing (Opus 5.5 at $4 in and $20 out per million tokens). Anthropic's models overview (its API ID, claude-opus-5-5). Read .
Claude Opus 5.5: what it means for you
Can I use Claude Opus 5.5 in llmwise?
Yes, since September 25, 2026. Claude Opus 5.5's messages a month: up to 31 on Plus ($10), 62 on Pro ($20), 200 on Max ($50), 450 on Ultra ($100), 1,000 on Studio ($200). It isn't in the free trial.
Claude Opus 5.5 is in llmwise
Up to 62 Claude Opus 5.5 messages a month on Pro ($20). Its count shows before you send.