
Claude Opus 5.1: The Opus Version Anthropic Never Shipped
- openaiNEWOpenAI: GPT-6.1 Sol2026-09-2952Intelligence
- anthropicNEWAnthropic: Claude Sonnet 5.52026-09-2856Intelligence
- typesafeNEWTypeSafe: Jev 1.132026-09-24$0.04 / $0.00 per 1M tokens · 145 tok/s
- OpenAINEWOpenAI: GPT-6 Luna2026-09-2238Intelligence
- OpenAINEWOpenAI: GPT-6 Sol2026-09-2248Intelligence
- AnthropicNEWAnthropic: Claude Opus 5.52026-09-2258Intelligence
- xAIGrok 4.72026-09-2146Intelligence
- OrcaOrca: OrcaCyber Zero 1.02026-09-17$3.00 / $7.50 per 1M tokens · 127 tok/s
- OrcaOrca: OrcaVerify Text 1.02026-09-16$2.00 / $0.00 per 1M tokens · 933 tok/s
- DeepSeekDeepSeek: DeepSeek V4.1 Flash2026-09-1040Intelligence
- OpenAIOpenAI: GPT-6 Astra2026-09-0453Intelligence77Coding
- GoogleGoogle: Gemini 3.8 Flash2026-09-0241Intelligence76Coding
- AlibabaQwen: Qwen3.8 Max (0902)2026-09-0245Intelligence76Coding
- AnthropicAnthropic: Claude Fable 5.12026-09-0153Intelligence82Coding
- TencentTencent: Hy4 preview2026-08-28$0.83 / $2.50 per 1M tokens · 52 tok/s
- AlibabaQwen: Qwen3.8 Flash2026-08-26$0.15 / $0.47 per 1M tokens · 183 tok/s
- z-aiZ.ai: GLM 5.3 Flash2026-08-2642Intelligence72Coding
- DeepSeekDeepSeek: DeepSeek V4 Flash Vision (Exp)2026-08-21$0.22 / $0.66 per 1M tokens · 217 tok/s
- z-aiZ.ai: GLM 5.32026-08-1845Intelligence75Coding
- obsidianQwen3.8 27B2026-08-1534Intelligence68Coding
If you searched for Claude Opus 5.1, the honest answer arrives in the first line: there is no model by that name to call. Claude Opus 5.1 does not appear in the vendor's model list, has no model ID, no price, no context window, no benchmark and no page of its own, and it is not a route on OrcaRouter. One character away sits Claude Opus 5.5, released September 22, 2026, which is the Opus model the vendor sells today; beside it sits Claude Opus 5, released July 24, 2026, which Opus 5.5 has superseded. Two of the three strings a reader might type are real models a single character apart. This page is the reference for the third one: what Claude Opus 5.1 is, why it was never released, and which Opus to call instead.
It is worth being explicit about what kind of page this is, because the version it documents has no launch to report. This is the canonical reference page for a model string that people search for — the page a reader reaches after typing the model's own name — not a news piece about a dated event. Its warrant is standing search demand, measured first-party: over the 28 days to September 25, 2026, the exact query "claude opus 5.1" was typed 2,235 times in Google Search, earning 93 clicks at a 4.16% click-through rate from an average position of 8.0, with no dedicated page behind any of it. The largest single loose query alongside the bare name is "claude opus 5.1 release date" (253 impressions, 22 clicks at 8.7%, position 7.3), which is why the release-date question is answered below rather than deferred. The one datable anchor is Opus 5.5's September 22 release, and it is used here as a fact that dates the Opus line, not as an event to report — our coverage of that launch already exists and this page is not a second version of it.
Which Opus you are actually looking for

The suffix confusion is the entire reason this page needs to exist, so start with the map. Anthropic's own model overview, read on September 28, 2026, lists four current models: Claude Fable 5.1, Claude Opus 5.5, Claude Sonnet 5 and Claude Haiku 4.5. Beneath those, in the same page's legacy list, sit Claude Fable 5, Claude Opus 5, Claude Opus 4.8, Claude Opus 4.7, Claude Opus 4.6, Claude Opus 4.5, Claude Sonnet 4.6 and Claude Sonnet 4.5. There is no fifth member of the Opus line in either list.
Three strings are in play, and only two of them resolve to anything:
• Claude Opus 5.1 — no entry anywhere on Anthropic's pages, no route in our catalogue, no rate card. This page.
• Claude Opus 5.5 — the current Opus. Released September 22, 2026, at $4 input / $20 output per million tokens. Anthropic's retirement commitment runs to not sooner than September 22, 2027, and it is the model the vendor's own migration guide tells Opus 5 users to move to.
• Claude Opus 5 — the previous Opus. Released July 24, 2026, at $5 / $25, still Active on Anthropic's lifecycle table with a commitment to not sooner than July 24, 2027, but no longer the flagship.
There is a fourth trap worth naming, because it is the one that actually has a paper trail. Claude Opus 4.1 is a real Anthropic model: it appears in the vendor's deprecation history as retired on August 5, 2026, with Claude Opus 4.8 as the recommended replacement. So "Opus 4.1" is a string with a documented release, a documented retirement date and a documented migration target, and "Opus 5.1" has none of those things. A reader who half-remembers an Opus .1 is most likely remembering 4.1, not a 5.1.
The .1 suffix itself is not imaginary, which is what keeps the confusion alive. Claude Fable 5 and Claude Mythos 5 got their .1 point releases on September 1, 2026, at the same $10 / $50 as the originals. Anthropic shipped two .1 models that day, and neither of them was an Opus.
The release date, answered directly
There is no release date for Claude Opus 5.1, because the version was never released. What readers are dating when they search this query is almost always one of two other things.
The first is Claude Opus 5.5, released September 22, 2026. That is the date on Anthropic's release notes and on its announcement page, and it is the date of the Opus release that actually happened in September.
The second is an expected window that produced no model. Through August 2026, community trackers attached rumoured dates to a possible Opus 5.1 — August 28 first, then a slip to "next week," then September 5 — and a Bedrock probe at the end of August produced a 404 that some read as the name being staged on AWS. None of it was vendor-confirmed, and none of it produced a launch; we documented that waiting period in the Claude Fable 5.1 & Opus 5.1 delay-leak post at the time, including the probe's own caveats. The window closed on September 1, when Anthropic's top-tier refresh turned out to be Claude Fable 5.1 and Claude Mythos 5.1 rather than an Opus point release at all. A single community account later reported, on September 14, that Claude Code was routing "opus" traffic to an unreleased checkpoint it labelled Claude Opus 5.2; that report is unverified, no identifier was confirmed, and it does not change the fact that nothing named Opus 5.1 was ever served.
The rest of the Opus line's real dates, for the record, all from Anthropic's release notes:
• Claude Opus 4.5 — November 24, 2025
• Claude Opus 4.6 — February 5, 2026
• Claude Opus 4.7 — April 16, 2026
• Claude Opus 4.8 — May 28, 2026
• Claude Opus 5 — July 24, 2026
• Claude Opus 5.5 — September 22, 2026
Between July 24 and September 22 the only Opus-tier movement was the flagship change itself. A five-week gap with no Opus .1 in it is the whole answer.
The envelope
There is no published envelope for Claude Opus 5.1 — no context window, no output limit, no modality list, no knowledge cutoff, because there is no model card to read them from. Rather than guess at what an Opus .1 would have carried, here is what the two Opus models that exist actually specify, all of it from Anthropic's own documentation.
• Context — 1M tokens on both Claude Opus 5 and Claude Opus 5.5, unchanged across the line
• Max output — 128K tokens synchronously on both; up to 300K on the Message Batches API behind the output-300k-2026-03-24 beta header
• Input and output — text, image and PDF/file input with text output, vision and tool use on every current model, per the vendor's models overview; none of the current lineup is text-only
• Reliable knowledge cutoff — June 2026 on both Opus 5 and Opus 5.5
• Effort — Claude Opus 5 defaults to high with a ladder running low, medium, high, xhigh, max; Claude Opus 5.5 defaults to medium with the same ladder
• Thinking — adaptive on both, but Opus 5 accepts thinking: {"type": "disabled"} at effort high or below while Claude Opus 5.5 always thinks and returns a 400 on that field
• Retirement commitment — not sooner than July 24, 2027 for Claude Opus 5; not sooner than September 22, 2027 for Claude Opus 5.5
The one envelope change between the two is a doubling-down on thinking: Opus 5.5 makes always-on adaptive thinking mandatory, which is a documented breaking change for any integration that had it switched off. That is the kind of line an Opus 5.1 page would have had to draw somewhere, and it is exactly the line nobody can draw for a model that was never published.
The rate card, in full, as it stands today
Claude Opus 5.1 has no rate card. Nothing on Anthropic's pricing page refers to it, and the two real Opus models price like this, read from that page on September 28, 2026:
• Claude Opus 5.5 — $4 input / $20 output per million tokens; 5-minute cache write $5; 1-hour cache write $8; cache read $0.20; Batch API $2 / $10
• Claude Opus 5 — $5 input / $25 output per million tokens; 5-minute cache write $6.25; 1-hour cache write $10; cache read $0.50; Batch API $2.50 / $12.50
• Fast mode, Claude API only and flagged by Anthropic as a research preview — Claude Opus 5.5 at $8 / $40, Claude Opus 5 at $10 / $50
Two things about this card are worth stating plainly, because both are easy to get wrong from a blog post. First, the rates have not moved since release: Claude Opus 5 has carried $5 / $25 since July 24, 2026, and Opus 5.5 arrived on September 22 already below it rather than being cut later. Second — and this is the line item most coverage skips — cache reads are priced as a multiplier of the base input rate, 0.1× as standard and 0.05× on Claude Opus 5.5. For an agent that re-sends a long system prompt and a file tree on every turn, that is the dominant cost, and it is why the two Opus models are further apart in practice than $5 versus $4 suggests.
What Anthropic documents about effort and extended thinking
For Claude Opus 5.1 there is nothing to document, and the absence is the finding. The vocabulary a reader probably wants anyway is on the vendor's current pages, and it is worth unpacking because "extended thinking" is now the wrong term for anything in this family.
Anthropic's model overview defines two regimes. Adaptive thinking lets the model decide how much to think, steered by the effort parameter inside output_config, with the ladder running low, medium, high, xhigh and max. Extended thinking is the older manual mode — thinking: {"type": "enabled", "budget_tokens": N} — and the same page states that it is deprecated on Claude Opus 4.6 and Claude Sonnet 4.6 and not accepted on later models. So on every Opus from 4.7 onward, effort is the control and a token budget is not.
That is the practical shape of an Anthropic point release, and the one live example is instructive. Between Claude Fable 5 and Claude Fable 5.1 the headline rate did not change — $10 / $50 both before and after — but the cache read fell from $1.00 to $0.25 per million tokens. A .1 at this vendor has meant a new pinned model ID, the same tier, and a cheaper run, not a new SKU wedged between two existing prices. If an Opus 5.1 had shipped, that is the pattern it would most plausibly have followed; it did not ship, and the pattern is offered here as the vendor's convention rather than as a prediction about a model.
Is it still current? No — and the thing it would have replaced is not either
Claude Opus 5.1 is not current, and it never was. No superseded-status row exists for it because no lifecycle row exists for it at all.
The two Opus models that do have rows are both labelled Active on Anthropic's deprecation table — but only one of them is the flagship. Claude Opus 5.5 is the current Opus: it is the model the vendor's choosing-a-model page recommends starting with, it is priced below Opus 5 on every line, and Anthropic publishes a migration guide whose title is, literally, migrating from Claude Opus 5. Claude Opus 5 remains functional and will keep working well into 2027, so an integration pinned to it is not broken and does not need an emergency change. It is simply no longer the top of the line, and the string to swap when you do move is claude-opus-5 to claude-opus-5-5.
For a reader who landed here by searching a version number that does not exist, that is the decision-relevant answer: there is nothing to migrate off, because you were never on it. Whatever you are running, it is almost certainly one of those two.
Availability on OrcaRouter

Verified against our live model list on September 28, 2026. Claude Opus 5.5 is one of our routes at anthropic/claude-opus-5.5, at $4.00 input / $20.00 output per million tokens, cache read $0.20 and cache write $5.00, with a 1M-token context window, 128K max output, a release date of 2026-09-22, and neither delisted nor marked upcoming. Claude Opus 5 is also live at anthropic/claude-opus-5, at $5.00 / $25.00 with cache read $0.50 and a 1M-token context, released 2026-07-24.

Claude Opus 5.1 is not. The same model endpoint that answers for both Opus models above returns "model not found" for anthropic/claude-opus-5-1, and no Opus 5.1 route exists to link. That is the honest availability answer, and it is the one a reader on a retired or phantom version most needs.
Both real Opus routes bill at Anthropic's published list rates with 0% markup — the provider's price passed through, so a vendor rate change is live here the same day rather than whenever a reseller refreshes a table. Both Opus routes live on one key, which matters for the swap above: moving from claude-opus-5 to claude-opus-5-5 is a string change, not a second contract, a second SDK or a second billing relationship. We also serve the rest of the current Claude line — Claude Fable 5.1, Claude Sonnet 5 and Claude Haiku 4.5 — alongside them on the same endpoint. If you want to compare the two Opus models on your own traffic rather than on a vendor chart, automatic failover is the cheap way to run the new one on a share of real requests without betting a production path on it.
How to check any "Claude Opus 5.1" claim you meet
Because the version number is in circulation, a short test is worth keeping. Anthropic's model IDs are pinned snapshots, and the vendor's own versioning page says an updated model ships under a new model ID rather than as an update to an existing one. So a version that exists leaves three fingerprints, and a version that does not leaves none of them:
• A model ID. It appears in the Models API and in the vendor's model list. A name with no ID is a name, not a model.
• A pricing row. Every Claude model that can be called has a row on Anthropic's pricing page with an input and an output rate.
• A lifecycle row. Deprecation and retirement commitments are published per model ID, so a manufactured version number typically has neither a release note nor a retirement date.
The same test works on our side: a model we serve has a page and a price in our catalogue, and a model we do not serve has neither. Applied to Claude Opus 5.1 today, all four checks come back empty — which is the whole content of this page. If Anthropic ships an Opus refresh, what changes is not a claim but a row: a new ID, a rate card and a lifecycle date. Until one of those appears, the search that brought you here has an answer, and the answer is that the model you are looking for was never made.
Compared in this article3
Detected from this article · Benchmarks: Artificial Analysis · updated daily
