33 events in a year. Then 155 in the next.
Every model-market event Vitna has recorded since Mar 2023, newest first. Each entry links to a primary source on a domain the provider itself publishes on, and carries the date that source was last fetched; the newest of those checks is 2026-11-19. The log only ever gains entries.
Every event at its true date, newest left.
Each move as a share of the price before it.
83 detected moves carry both of their periods, so each bar is the move as a share of the price before it: cuts run left, rises run right, and the longest bar is the largest move. Each model leads to its own page and each meter to the move's own page. Curated entries cite an announcement rather than two readings and list below without a bar.
Every entry, with the document behind it.
47 of 216 events. Anthropic.
- 2026Anthropicprice · detectedsource ↗
Anthropic Claude Sonnet 5.5 cached input price falls to $0.10
Cached input price for Anthropic Claude Sonnet 5.5 moves from $0.20 to $0.10 in USD per 1M tokens, a 50 percent cut. Read on 7 Oct 2026 at 7 Oct 22:00 UTC; the documents before and after are archived as 9c0140eac6b4 and 6b5f41edbdbd.
- Anthropiclaunchsource ↗
Claude Opus 5.5 released
Succeeds Opus 5 at $4 / $20, a list price 20 percent below Opus 5's $5 / $25, with cache reads at 0.05x base input rather than 0.1x.
- Anthropicpricesource ↗
Claude Sonnet 5 introductory price becomes the standard price
The $2 / $10 per million tokens announced at launch was to rise to $3 / $15 on 1 September. That increase will not occur.
- Anthropicpolicysource ↗
Claude Opus 4.1 retired
Corrected. The row that supersedes this one stands in the record.
Deprecated June 5 with 60 days' notice; requests to the model now fail. The recommended replacement is Opus 4.8.
- Anthropicavailabilitysource ↗
Claude Opus 4.1 is retired
A correction, superseding the row it replaces.
Corrects the kind on the row this supersedes, which was entered as policy: a retirement is an availability event, and it is how every other retirement in this record is filed. Deprecated 5 June with the sixty days of notice the policy promises; the recommended replacement is Opus 4.8. The dates on the cited page bind Anthropic-operated platforms, and Amazon Bedrock and Google Cloud set their own retirement schedules.
- Anthropiclaunchsource ↗
Claude Opus 5 released
Succeeds Opus 4.5 at the same $5 / $25 list price.
- Anthropiclaunchsource ↗
Claude Sonnet 5 released
Opus-class coding and tool use at a $2 / $10 list price.
- Anthropicavailabilitysource ↗
Claude Sonnet 4 and Claude Opus 4 are retired
Requests to both now fail on the Claude API. Each remains served on at least one partner cloud.
- Anthropicpolicysource ↗
US export-control directive suspends Fable 5 and Mythos 5
All customer access disabled three days after launch; restored at the end of June.
- Anthropiclaunchsource ↗
Claude Fable 5 opens the Mythos tier
The first publicly available model above the Opus class, at $10 / $50.
- Anthropicdeprecationsource ↗
Claude Opus 4.1 is deprecated
Retirement set for 5 August 2026, the sixty days of notice the policy promises.
- Anthropiclaunchsource ↗
Claude Opus 4.8 released
A 1M token context window by default across the Claude API, Bedrock, Google Cloud and Microsoft Foundry.
- Anthropicavailabilitysource ↗
Claude Haiku 3 is retired
Requests now fail, twenty-five months after the model launched with the Claude 3 family.
- Anthropiclaunchsource ↗
Claude Opus 4.7 released at the price of Opus 4.6
The same $5 / $25 per million tokens, on a new tokenizer that produces roughly 30 percent more tokens for the same text.
- Anthropicdeprecationsource ↗
Claude Sonnet 4 and Claude Opus 4 are deprecated
Retirement set for 15 June 2026 for both, sixty-two days of notice.
- Anthropicavailabilitysource ↗
Claude Mythos Preview opens by invitation only
A gated research preview for defensive cybersecurity work under Project Glasswing. There is no self-serve sign-up.
- Anthropicpolicysource ↗
The 1M token context window leaves beta at standard pricing
Requests over 200k tokens work with no beta header on Opus 4.6 and Sonnet 4.6, and are billed at the same per-token rate as a short one.
- Anthropicavailabilitysource ↗
Claude Sonnet 3.7 and Claude Haiku 3.5 are retired
Both stop answering on the Claude API on the same day. Haiku 3.5 remains served on Amazon Bedrock and Google Cloud.
- Anthropicdeprecationsource ↗
Claude Haiku 3 is deprecated
Retirement set for 20 April 2026, announced on the day two other models are retired.
- Anthropiclaunchsource ↗
Claude Sonnet 4.6 released
The balanced tier gains the 1M token context window and better agentic search at the same $3 / $15.
- Anthropicpolicysource ↗
US-only inference is priced at a premium
Data residency controls arrive with the inference_geo parameter; pinning inference to the US costs 1.1x standard rates on models released after 1 February 2026.
- Anthropiclaunchsource ↗
Claude Opus 4.6 released
Recommends adaptive thinking, and opens the 1M token context window in beta on the Opus line.
- Anthropicavailabilitysource ↗
Claude Opus 3 is retired
Requests now fail, more than six months after the notice and twenty-two months after the model launched.
- 2025Anthropicdeprecationsource ↗
Claude Haiku 3.5 is deprecated
Retirement set for 19 February 2026, the same day Sonnet 3.7 goes.
- Anthropiclaunchsource ↗
Claude Opus 4.5 released
The frontier tier moves to $5 / $25 per million tokens.
- Anthropicavailabilitysource ↗
The Claude Sonnet 3.5 snapshots are retired
Requests to both now fail, seventy-six days after the notice.
- Anthropicdeprecationsource ↗
Claude Sonnet 3.7 is deprecated on the day 3.5 goes
Retirement set for 19 February 2026, announced in the same release notes that retire its predecessor.
- Anthropiclaunchsource ↗
Claude Haiku 4.5 released
Near-frontier small-model quality at $1 / $5 per million tokens.
- Anthropicpolicysource ↗
Global endpoint pricing introduced on Bedrock and Vertex AI
Regional routing starts carrying a premium over global on the partner clouds. First-party Claude API pricing is unaffected.
- Anthropiclaunchsource ↗
Claude Sonnet 4.5 released
The balanced tier is refreshed for agents and coding at $3 / $15 per million tokens.
- Anthropicdeprecationsource ↗
Both Claude Sonnet 3.5 snapshots are deprecated
Retirement set for 28 October 2025 for the June and October 2024 snapshots.
- Anthropiclaunchsource ↗
Claude Opus 4.1 released
An incremental update to Opus 4 at the same frontier list price.
- Anthropicavailabilitysource ↗
Claude 2.0, Claude 2.1 and Claude Sonnet 3 are retired
Requests to all three now fail, on the date named in January.
- Anthropicdeprecationsource ↗
Claude Opus 3 is deprecated
Retirement set for 5 January 2026, more than six months of notice.
- Anthropiclaunchsource ↗
Claude Opus 4 and Claude Sonnet 4 launch together
The generation that introduces extended thinking across both tiers.
- Anthropiclaunchsource ↗
Claude Sonnet 3.7 ships one model with two response modes
Near-instant answers or step-by-step extended thinking, selected per request rather than by choosing a model.
- Anthropicdeprecationsource ↗
Claude 2, Claude 2.1 and Claude Sonnet 3 are deprecated
Retirement set for 21 July 2025, six months of notice for three models at once.
- 2024Anthropicavailabilitysource ↗
Claude 1 and the Instant models are retired
Requests to claude-1.0 through claude-instant-1.2 now fail, sixty-three days after the notice.
- Anthropiclaunchsource ↗
Claude Haiku 3.5 arrives as a text-only model
The fast tier is refreshed on the Claude API without vision at launch.
- Anthropiclaunchsource ↗
Claude Sonnet 3.5 is upgraded and gains computer use
A second Sonnet 3.5 snapshot ships with Anthropic-defined computer use tools.
- Anthropicpolicysource ↗
The Message Batches API opens at half the standard rate
Large asynchronous batches are processed for 50 percent less, matching the discount OpenAI set in April.
- Anthropicdeprecationsource ↗
Claude 1 and the Instant models are deprecated
Retirement set for 6 November 2024. The first deprecation notice the record carries.
- Anthropicpolicysource ↗
Prompt caching enters beta on the Claude API
A cached prompt prefix is re-read at a fraction of the input rate, which changes the unit price of a repeated system prompt rather than the model.
- Anthropiclaunchsource ↗
Claude Sonnet 3.5 lands across three channels at once
Available on the Claude API, Amazon Bedrock and Vertex AI on the day it is announced.
- Anthropiclaunchsource ↗
The Claude 3 family sets three price tiers at once
Opus, Sonnet and Haiku launch together at $15 / $75, $3 / $15 and $0.25 / $1.25 per million tokens, the first time one announcement names a whole ladder.
- 2023Anthropiclaunchsource ↗
Claude 2 released at the price of Claude 1.3
A generation that improves on coding, maths and reasoning and does not move the rate.
- Anthropiclaunchsource ↗
Claude and Claude Instant open the Anthropic API
Two versions on the first day: a high-performance model, and a lighter, faster, cheaper one. The oldest event the record can primary-source.
An event appears here only with a primary-source URL verified on the date the entry carries. Secondary coverage can corroborate a date but never stands as the source. A detected price move is published only after a second read at least an hour later agrees, and cites the archived document either side of it.
A week with 3 events or more is a page of its own, with its counts, its sized moves and every entry in it. The newest is the week of 16 Nov 2026, with 16 price moves, and all 16 are listed by week.
This log is published as a JSON Feed at /api/v1/feed.json, public and with no key, carrying every entry here and the decisions the engine has proposed against them, and as RSS at /log/rss.xml for a feed reader, with the same items under the same ids. Every headline above is also a page of its own, at an address that keeps pointing at it. Nothing in either feed is a recommendation to act.
