The generally available Claude line runs from Claude and Instant on 2023-03-14 through Fable 5.1 on 2026-09-01: 2 (2023-07-11), 2.1 (2023-11-21), 3 (2024-03-04), 3.5 Sonnet (2024-06-21), 3.5 Haiku (2024-10-22), 3.7 (2025-02-24), 4 (2025-05-22), Opus 4.1 (2025-08-05), Sonnet 4.5 (2025-09-29), Haiku 4.5 (2025-10-15), Opus 4.5 (2025-11-24), Opus 4.6 (2026-02-05), Sonnet 4.6 (2026-02-17), Opus 4.7 (2026-04-16), Opus 4.8 (2026-05-28), Fable 5 (2026-06-09), Sonnet 5 (2026-06-30), Opus 5 (2026-07-24), Fable 5.1 (2026-09-01). The Scrubber plays each date as a frozen frame: what was true that day, sourced, and not rewritten after the fact. Claude Archive is an independent educational publication and is not affiliated with Anthropic. Live current rates belong at Claude Reports and Claude Central; weekly reaction belongs at Claude Weekly.
Verified September 19, 202620 GA launches24 frozen framesClaude Archive is an independent educational publication and is not affiliated with Anthropic.
Anthropic introduced Claude more broadly after a closed alpha with partners, offering Claude and Claude Instant through chat and API surfaces. This is the first public launch date in the archive. There is no Haiku / Sonnet / Opus ladder yet, and no later family name should be read back onto this day.
Claude 2 arrived with API access, claude.ai beta access in the US and UK, and a 100K-token input window. The launch post said the API was offered at the same price as Claude 1.3. That is a same-price note, not a dollar table, and this frame keeps it that way.
True that day
Claude 2 in public beta. Claude 1 and Instant remain in the historical record.
Claude 2.1 added a 200K-token context window, system prompts, beta tool use, and an updated pricing schedule described as improving cost efficiency across models. The 200K window is the headline fact of this day. System prompts here are a product feature announcement, not a published prompt archive.
Claude 3 debuted as Haiku, Sonnet, and Opus. Those three names become the persistent speed-to-capability ladder for later releases. This frame records the family structure arriving, not any later 4.x or 5-era model.
Anthropic launched the first Claude 3.5 model at the Sonnet tier with a 200K context window and $3 input / $15 output per million tokens. That rate is a launch-day fact. It is not a live quote for 2026.
True that day
Claude 3.5 Sonnet is the new Sonnet-class model alongside the Claude 3 family.
Anthropic announced an upgraded Claude 3.5 Sonnet, Claude 3.5 Haiku, and public beta computer use. Computer use is a capability announcement on this date, not a how-to. The upgraded Sonnet 3.5 and Haiku 3.5 are the new generally available 3.5-era models.
True that day
Upgraded Claude 3.5 Sonnet, Claude 3.5 Haiku, with Claude 3 Opus still in the family.
Claude 3.7 Sonnet introduced hybrid reasoning and extended thinking. Claude Code appeared as a limited research preview the same day. This archive records the date. Claude Code practice belongs to Claude Ships Code, not this frame.
True that day
Claude 3.7 Sonnet is the new Sonnet-class model. Claude Code is a research preview.
Claude Opus 4 and Claude Sonnet 4 launched with stronger coding, agent, and reasoning claims. Claude Code became generally available the same day. The 4-era names start here. No later 4.5, 4.6, 4.7, or 4.8 snapshot is present on this date.
True that day
Claude Opus 4, Claude Sonnet 4. Claude Code generally available.
Opus 4.1 upgraded Opus 4 for agentic tasks, real-world coding, and reasoning while keeping Opus 4 pricing. Same-price relative to Opus 4 is the sourced claim. This is a launch, not a retirement.
True that day
Claude Opus 4.1, Claude Sonnet 4, with earlier 3.x models still in the ledger until their own dates.
Sonnet 4.5 launched with coding, agent, computer-use, and reasoning gains and retained Sonnet 4 pricing at $3 input / $15 output per million tokens. That $3/$15 is the Sonnet 4.5 launch rate. It is not Sonnet 5, and it is not a 2026 live desk.
Haiku 4.5 launched as the speed and cost-efficiency member of the modern family. The launch post positions it as near-frontier, faster, and cheaper. This frame does not invent a dollar table the launch post does not carry here.
True that day
Claude Haiku 4.5, Claude Sonnet 4.5, Claude Opus 4.1.
Opus 4.5 launched on apps, API, and major cloud platforms at $5 input / $25 output per million tokens. That $5/$25 reset is the modern Opus rate beginning on this date. Later Opus models that keep the same rate do not rewrite this launch.
True that day
Claude Opus 4.5, Claude Sonnet 4.5, Claude Haiku 4.5.
Opus 4.6 launched with a 1M-token context-window beta for the Opus class. The 1M window is a beta on this date, not yet the later default listed for 5-era models. Pricing stays with the Opus 4.5 $5/$25 band; this frame does not invent a new rate.
True that day
Claude Opus 4.6, Claude Sonnet 4.5, Claude Haiku 4.5.
Sonnet 4.6 became the default model for Free and Pro users and added a 1M-token context-window beta. Same pricing as Sonnet 4.5. This is the last Sonnet-class launch before Sonnet 5.
True that day
Claude Sonnet 4.6, Claude Opus 4.6, Claude Haiku 4.5.
Opus 4.7 launched with new cybersecurity safeguards and the same $5/$25 per million tokens as Opus 4.6. Safeguards are recorded as a launch fact. This archive does not turn them into a bypass guide.
True that day
Claude Opus 4.7, Claude Sonnet 4.6, Claude Haiku 4.5.
Opus 4.8 launched with effort controls, dynamic workflows, and cheaper fast mode than prior Opus fast mode. On this date it is the current Opus. The later Opus 5 launch does not get written back onto this frame.
True that day
Claude Opus 4.8, Claude Sonnet 4.6, Claude Haiku 4.5.
Anthropic launched Fable 5 for general use and Mythos 5 for limited approved cybersecurity access. On this date they are new. The later suspension is not part of this frame. Fable 5.1 does not exist yet.
True that day
Fable 5 generally available. Mythos 5 limited access. Opus 4.8, Sonnet 4.6, Haiku 4.5 still current in their tiers.
Anthropic published "Redeploying Fable 5", setting out the suspension of Fable 5 and Mythos 5 and the restoration of access starting July 1, 2026. The June 9 launch is not deleted by this post. The archive keeps launch, suspension, and return as three dated events.
True that day
Fable 5 and Mythos 5 are suspended on this date, with restoration described as starting 2026-07-01.
Sonnet 5 launched as the newest Sonnet model at $2 input / $10 output per million tokens, announced as introductory pricing through August 31, 2026, with a planned standard rate of $3/$15 to follow. That is what the launch post said on this day. This frame does not replace that sentence with the later non-increase.
True that day
Claude Sonnet 5 is the new Sonnet-class model. Opus 4.8 remains the current Opus on this date. Fable 5 is in the redeployment sequence described the same day.
Dated rate
$2 / $10 per MTok, announced as introductory through 2026-08-31.
Opus 5 launched as the new Opus-class model at the same $5/$25 per million tokens as Opus 4.8, became the default on Claude Max, and added a fast mode at roughly 2.5x speed for 2x the base price. Model ID claude-opus-5. 1M context. On this date, "Opus 4.8 is the current Opus" stops being true going forward. This frame does not rewrite earlier Opus 4.8 frames.
True that day
Claude Opus 5, Claude Sonnet 5, Claude Haiku 4.5. Fable 5 remains in the 5-era record; Fable 5.1 has not launched.
Dated rate
$5 / $25 per MTok. Fast mode 2x base for ~2.5x speed.
Claude Opus 4.1 retired on schedule on August 5, 2026, closing the deadline opened by the June 5, 2026 deprecation notice. Requests to claude-opus-4-1-20250805 fail on Anthropic-operated platforms. The named replacement in that notice was claude-opus-4-8, a string written when the notice was published.
True that day
Active Opus-class model on this date is Claude Opus 5, which launched 2026-07-24. Opus 4.1 is retired.
Anthropic announced that future Claude models watermark generated text for EU AI Act and Code of Practice transparency obligations. The watermark is global at launch, adds no tokens and no hidden characters, and is not traceable to a person, organization, or conversation, with a transition period for models released before August 2, 2026. This is a dated policy event in the model timeline, not an everyday how-to.
True that day
Same generally available lineup as the prior frame: Opus 5, Sonnet 5, Haiku 4.5, with Fable 5 still in the 5-era history.
The $2/$10 per million-token rate announced as introductory through August 31, 2026 did not rise to $3/$15 on September 1. The pricing docs carry $2/$10 as the standard Sonnet 5 rate. This is a dated price event: a planned increase that did not occur. The June 30 launch frame still says the rate was introductory, because that is what was true that day.
True that day
Claude Sonnet 5 remains the Sonnet-class model. This event changes the rate card, not the model ID.
Dated rate
$2 / $10 per MTok is now the standard Sonnet 5 rate. The $3/$15 plan is superseded.
Live current rates are not this archive. Use Claude Reports for analysis and Claude Central plans and pricing for the current card. This frame only records that the announced Sonnet 5 rise did not happen.
Fable 5.1 launched as the most capable generally available Claude model at $10/$50 per million tokens, with cache reads cut to $0.25 per million tokens (2.5% of base input). Model ID claude-fable-5-1. 1M context. Mythos 5.1 shares the underlying model under trusted access only. The June Fable 5 launch, suspension, and redeployment remain in the film; they are not deleted because a successor arrived.
True that day
Claude Fable 5.1, Claude Opus 5, Claude Sonnet 5, Claude Haiku 4.5. Mythos 5.1 trusted access only.
Drag the slider, click a sprocket, or use arrow keys when the gate has focus. Space plays and pauses. L toggles the later overlay. The frozen text never changes.
How to read the film
A Claude launch page stays a launch page after the model retires. A planned price increase stays planned until a later dated source says it did not happen. The Scrubber is built so that discipline is visible: twenty-four frames, from Claude and Instant on March 14, 2023 through Fable 5.1 on September 1, 2026, each frozen to its own day.
The June Fable sequence is three frames, not one. Fable 5 and Mythos 5 launched June 9, 2026. The redeployment post on June 30 recorded the suspension and the return starting July 1. Fable 5.1 launched September 1 and superseded Fable 5 as the most capable generally available model. Deleting the first two would be a rewrite. The film keeps all three.
The Sonnet 5 rate is two frames, not a live quote. On June 30, 2026 the $2/$10 rate was announced as introductory through August 31 with $3/$15 to follow. On September 1 the increase did not happen. The second frame points at Claude Reports and Claude Central for live current rates. This site records the event.
Opus 5 on July 24, 2026 is a launch, not a backdated correction. Earlier Opus 4.8 frames still say Opus 4.8 launched as the current Opus on May 28. The later overlay, if you turn it on, can name the successor. The frozen body cannot.
System-prompt history on this site is frozen at the July 6, 2026 verification, covering published web and app prompt dates through Claude Fable 5. No Opus 5 or Fable 5.1 prompt text is invented or pasted here. See the system-prompt guide and Anthropic's official release notes. Weekly reaction belongs at Claude Weekly. The table form of this film is the timeline.
What is the complete Claude model timeline from Claude 1 to Claude Fable 5.1?
Generally available dates: Claude and Instant on 2023-03-14; Claude 2 on 2023-07-11; Claude 2.1 on 2023-11-21; Claude 3 Haiku, Sonnet, and Opus on 2024-03-04; Claude 3.5 Sonnet on 2024-06-21; Claude 3.5 Haiku on 2024-10-22; Claude 3.7 Sonnet on 2025-02-24; Claude 4 (Opus 4 and Sonnet 4) on 2025-05-22; Opus 4.1 on 2025-08-05; Sonnet 4.5 on 2025-09-29; Haiku 4.5 on 2025-10-15; Opus 4.5 on 2025-11-24; Opus 4.6 on 2026-02-05; Sonnet 4.6 on 2026-02-17; Opus 4.7 on 2026-04-16; Opus 4.8 on 2026-05-28; Fable 5 and Mythos 5 on 2026-06-09; Sonnet 5 on 2026-06-30; Opus 5 on 2026-07-24; Fable 5.1 and Mythos 5.1 on 2026-09-01. The Scrubber plays each date as a frozen frame that is not rewritten after the fact.
Did Claude Sonnet 5 pricing increase on September 1, 2026?
No. The $2/$10 per MTok rate announced as introductory through August 31, 2026 did not rise to $3/$15. The archive records that non-increase as a dated event on 2026-09-01. Live current rates belong at claudereports.com and claudecentral.com, not on this historical desk.
When did Claude Opus 5 and Claude Fable 5.1 launch?
Claude Opus 5 launched on July 24, 2026 at $5/$25 per MTok. Claude Fable 5.1 and Claude Mythos 5.1 launched on September 1, 2026. Fable 5.1 is generally available at $10/$50 per MTok with cache reads at $0.25 per MTok; Mythos 5.1 is trusted access only.
Does The Scrubber rewrite older Claude launch pages when a newer model ships?
No. Each frame is what was true that day. The June 9, 2026 Fable 5 launch stays a launch even after the suspension, the June 30 redeployment post, and the September 1 Fable 5.1 successor. The June 30 Sonnet 5 frame still records introductory pricing through August 31, and the September 1 frame records that the increase did not happen.
Are Claude Opus 5 or Claude Fable 5.1 system prompts listed here?
No. The system-prompt table on this site is verified through the July 6, 2026 source pass and covers models up to Claude Fable 5. The archive does not invent later prompt leaks or paste proprietary prompts. Check Anthropic's official system-prompt release notes for anything after that freeze.
Is Claude Archive affiliated with Anthropic?
No. Claude Archive is an independent educational publication and is not affiliated with Anthropic.