Claude Code changelog

Claude Code makes Haiku 5.5 the default Haiku model

Claude Code 2.1.293 added Claude Haiku 5.5 as the default Haiku model on the Anthropic API.

Opus 5.5 in Claude Code tops the official Terminal-Bench 4.0 leaderboard at 64.8%

Claude Code with Opus 5.5 at max effort resolved 64.85% of Terminal-Bench 4.0 tasks (plus or minus 3.1, 330 trials, $4,718.65 total cost), ranked first. The entry was posted October 6, 2026.

Sonnet 5.5 in Claude Code scores 61.8% on Terminal-Bench 4.0

Claude Code with Sonnet 5.5 at max effort resolved 61.82% (plus or minus 2.9, 330 trials, $7,333.64 total cost), second on the leaderboard. Posted October 6, 2026.

GitHub issue claims Opus 5.5 quality fell back to Opus 5 level

A single Claude Code GitHub issue on version 2.1.284 claims Opus 5.5 quality degraded to the level of Opus 5. It was labeled needs-info and has no Hacker News thread. Livenerf found no change in the same period.

Livenerf starts a 30-day test of whether Opus 5.5 gets worse after launch

Livenerf runs a frozen 78-question panel (from GPQA Diamond, MMLU-Pro, competition math and AIME) through headless Claude Code on a Max plan, 90 samples a day for 30 days starting September 24, 2026, with a pinned CLI and exact-match grading. The Hacker News launch thread drew 926 points and 394 comments.

Who is affected:

Claude Max subscribers who suspect Opus 5.5 has been quietly downgraded.

What to do:

No action needed. Watch for the first decision around October 24, 2026.

MarginLab Claude Code tracker collects an Opus 5.5 baseline, with detection paused

As of its September 29, 2026 update, MarginLab's Claude Code tracker (Opus 5.5 at high effort, from September 24) showed a 77% daily pass rate on 50 tasks, 76% over 7 days (350 tasks) and 81% over 30 days (1,200 tasks). Degradation detection is paused until a baseline exists.

Who is affected:

People watching Claude Code with Opus 5.5 for silent regressions.

What to do:

No action needed. Check the tracker again once the baseline is set.

Claude Code makes Sonnet 5.5 the default Sonnet model

Claude Code 2.1.284 added Claude Sonnet 5.5 as the default Sonnet model on the Anthropic API.

Claude Code defaults to Opus 5.5, including on Pro plans

Claude Code 2.1.280 made Opus 5.5 the default Opus model and changed the default on Pro and Team Standard plans from Sonnet to Opus.

Who is affected:

Claude Code users on Pro and Team Standard plans, whose usage limits may now be consumed by Opus.

What to do:

Use /model to pick Sonnet if you want to stretch usage limits.

Five-hour usage limits raised with Opus 5.5, plus a saved limit reset

Alongside Opus 5.5, Anthropic raised five-hour usage limits on Pro, Max, Team and seat-based Enterprise plans and gave subscribers one rate limit reset they can use when they choose.

Opus 5 in Claude Code scores 53.9% on Terminal-Bench 4.0 at xhigh effort

Claude Code with Opus 5 scored 53.94% at xhigh, 51.82% at max (posted August 27), 50.3% at high, 44.85% at medium and 34.85% at low, each over 330 trials.

Who is affected:

Teams still on Opus 5 deciding whether to move to Opus 5.5.

What to do:

Compare against Opus 5.5 (64.85%) before upgrading.

Claude Code weekly limits settle 25% above pre-promotion levels

The 50% weekly limit promotion ended on September 13, 2026. From September 14, weekly Claude Code limits are 25% higher than before the promotion, which is lower than during it.

Who is affected:

Claude Code users on Pro, Max, Team and seat-based Enterprise plans.

Action needed:

Check /usage in Claude Code. Heavy users get less weekly headroom than during the promotion.

Claude Code makes Fable 5.1 the default Fable model

Claude Code 2.1.257 added Claude Fable 5.1 as the default Fable model with a 1M context window at $10 / $50 per million tokens.

Who is affected:

Claude Code users on the fable alias.

What to do:

Update Claude Code. Fable 5.1 needs version 2.1.255 or later.

Hacker News thread: Claude Code weekly limit to fall 17% compared with today

A Hacker News post linking to an X post by ClaudeDevs said Claude Code's weekly limit would drop 17% compared with the then-current level, drawing 67 points and 52 comments. This matches the end of the 50% promotion with a permanent 25% increase. The X post could not be opened.

Claude Code default model for seat-based Enterprise switches to Opus 5

Claude Code 2.1.251 changed the default model for seat-based Enterprise subscriptions to Opus 5, matching other premium plans.

Who is affected:

Claude Code users on seat-based Enterprise plans.

What to do:

Pick a model with /model if you want a different default.

Report: Claude Code 2.1.237 sent the 'high' effort setting as the old 'low' value

A MarsBit report on KuCoin says a developer found Claude Code 2.1.237 mapped the 'high' reasoning level to the value 10, previously used for 'low', as part of an undisclosed A/B test, and that an Anthropic engineer confirmed the issue but said it did not affect model performance.

Who is affected:

Claude Code users on version 2.1.237 who selected high effort.

What to do:

Update Claude Code and confirm the effort level with /effort.

Claude Code removes todo tools on newer models

Claude Code 2.1.233 removed the TodoWrite and task-tracking tools on Opus 4.8, Sonnet 5, Fable 5, Mythos 5 and newer models. An environment variable brings them back.

Who is affected:

Claude Code hooks, scripts or SDK apps that depend on TodoWrite or TaskCreate.

What to do:

Set CLAUDE_CODE_ENABLE_TODO_TOOLS=1 if you rely on those tools.

Claude Code makes Opus 5 the default Opus model

Claude Code 2.1.219 added Claude Opus 5 as the default Opus model and removed Opus 4.7 from fast mode. Anthropic also made Opus 5 the default model on Claude Max.

Who is affected:

Claude Code users on the opus alias or the Max plan.

What to do:

Pin claude-opus-4-8 with /model if you need the previous model.

Claude Sonnet 5 becomes the default model in Claude Code

Claude Code 2.1.197 made Claude Sonnet 5 the default model, with a native 1M token context window.

Who is affected:

Claude Code users who had not picked a model.

What to do:

Update to Claude Code 2.1.197 or later. Use /model to pick a different default.

Claude Code Max users default to fast mode on Opus 4.8

In Claude Code, Max plan users now default to fast mode when using Claude Opus 4.8.

Who is affected:

Claude Code users on the Max plan.

What to do:

Run /fast to turn fast mode off if you prefer standard speed.

MarginLab tracker measures a five-day Claude Code drop tied to CLI version 2.1.150

MarginLab's daily SWE-Bench-Pro subset run of Claude Code (50 tasks per day) fell from a 65% Opus 4.7 baseline to 50 to 54% from May 22 to May 26, 2026, and recovered to 66% when version 2.1.153 landed. Tool calls per task rose about 60% over the same days, and MarginLab concluded it was a harness issue, not a model regression.

Who is affected:

Claude Code users on CLI versions 2.1.150 to 2.1.152 in late May 2026.

What to do:

No action needed. Later CLI versions recovered.

Claude Code weekly limits raised 50% for a limited time

From May 13 through September 13, 2026, weekly usage limits in Claude Code were 50% higher for Pro, Max, Team and legacy seat-based Enterprise plans. Five-hour limits were unchanged.

Claude Code five-hour limits doubled and peak-hour cut removed

Anthropic doubled Claude Code five-hour rate limits for Pro, Max, Team and seat-based Enterprise plans, removed the peak-hours limit reduction for Pro and Max, and raised API rate limits for Opus models.

Report: Claude Code refuses requests or bills extra usage when commits mention OpenClaw

A post on X, discussed on Hacker News with 1,349 points and 720 comments, claimed Claude Code refuses requests or charges extra usage when a repository's commits mention OpenClaw. The X post could not be opened, so the claim is unverified.

Who is affected:

Claude Code subscribers working in repositories that reference OpenClaw.

What to do:

No action needed until Anthropic confirms the behavior.

Claude Code regression: malware reminder makes subagents refuse legitimate edits

A GitHub issue opened April 16, 2026 reported that Claude Code v2.1.111 still injects a malware system reminder into every file read, causing Opus 4.7 subagents to refuse code edits, after an earlier fix in v2.1.92. The thread drew 252 points and 147 comments.

Who is affected:

Claude Code users running subagents that edit code.

What to do:

Update Claude Code and report refusals with /feedback.

Anthropic confirms three changes degraded Claude Code quality

Anthropic traced user reports of worse Claude Code answers to a lower default effort (March 4 to April 7), a bug that dropped older reasoning every turn (March 26 to April 10) and a verbosity prompt change (April 16 to 20). The API was not affected, and Anthropic reset usage limits for all subscribers on April 23.

Who is affected:

Claude Code, Agent SDK and Cowork users on Sonnet 4.6, Opus 4.6 and Opus 4.7 between March and April 2026.

Action needed:

Update Claude Code to 2.1.116 or later.

Anthropic briefly removes Claude Code from the Pro plan for some new sign-ups

Anthropic's pricing pages briefly dropped Claude Code from the $20 Pro plan. Its head of growth called it a test on 2% of new prosumer sign-ups. The Hacker News threads drew 683 points and 642 comments, and 79 points and 38 comments. The current pricing page lists Claude Code on Pro.

Who is affected:

People signing up for Claude Pro to use Claude Code.

What to do:

No action needed. Pro currently includes Claude Code.

OpenClaw supports reusing a Claude Code login again after Anthropic allows CLI-style use

OpenClaw's Anthropic provider docs list a Claude CLI route that reuses an existing Claude Code login through the installed executable, alongside API key billing. The Hacker News thread, titled as Anthropic allowing this usage again, drew 511 points and 293 comments.

Who is affected:

OpenClaw users with a Claude subscription.

What to do:

Use the Claude CLI route in OpenClaw rather than raw OAuth tokens.

User analysis says Claude Code's prompt cache TTL dropped from 1 hour to 5 minutes in early March

An analysis of 119,866 API calls from January 11 to April 11, 2026 claimed the cache TTL silently changed from 1 hour to 5 minutes around early March, raising cache creation costs 20 to 32% and quota use. The Hacker News thread drew 552 points and 420 comments.

Who is affected:

Claude Code subscribers whose quota drained faster from March 2026.

What to do:

No action needed. See Anthropic's April 23 postmortem.

Max 5x users report quota exhausted in 1.5 hours, Claude Code team points to 1M-context cache misses

A GitHub issue reported a Max 5x quota exhausted within 1.5 hours of moderate use on Opus 4.6 with 1M context. Claude Code's Boris Cherny replied that stale sessions over an hour old cause full cache misses with the 1M window and that Anthropic was investigating a 400k default. The thread drew 759 points and 660 comments.

Who is affected:

Claude Max subscribers using 1M-context sessions in Claude Code.

Action needed:

Run /clear before resuming sessions idle for over an hour, or set CLAUDE_CODE_AUTO_COMPACT_WINDOW=400000.

AMD AI director reports Claude Code got dumber and lazier since February

The Register reported that AMD's AI group director analyzed 6,852 Claude Code sessions with 234,760 tool calls and concluded Claude Code could not be trusted for complex engineering tasks. Anthropic later attributed the period to three product changes.

Who is affected:

Teams running long, complex Claude Code sessions in February to April 2026.

What to do:

No action needed. See the April 23 postmortem.

Claude subscriptions no longer cover OpenClaw and other third-party harnesses

From April 4, 2026 at noon Pacific, Claude subscription limits no longer apply to third-party harnesses starting with OpenClaw. Using them with a Claude login requires pay-as-you-go extra usage. Anthropic offered a one-time credit equal to the monthly subscription price.

Who is affected:

Pro and Max subscribers who ran OpenClaw or other third-party harnesses on their Claude login.

Action needed:

Turn on extra usage or switch the harness to an API key.

Claude subscription limits stop covering OpenClaw and other third-party harnesses

Anthropic emailed subscribers that from April 4, 2026 at 12pm PT, third-party harnesses starting with OpenClaw require pay-as-you-go extra usage instead of subscription limits, with a one-time credit equal to the monthly price redeemable by April 17 and up to 30% off pre-purchased bundles. The Hacker News thread drew 1,099 points and 827 comments.

Who is affected:

Claude Pro and Max subscribers who run OpenClaw or other third-party agents on their Claude login.

Action needed:

Turn on extra usage or switch the third-party tool to an API key, and redeem the credit by April 17, 2026.

Anthropic acknowledges Claude Code users hitting usage limits far faster than expected

The Register reported high token usage and early quota exhaustion. Anthropic said it was investigating as the top priority, after a peak-hours quota reduction it said would affect about 7% of users and the end of a doubled-limits promotion. The Hacker News thread drew 331 points and 224 comments.

Who is affected:

Claude Pro and Max subscribers using Claude Code in late March 2026.

Action needed:

Update Claude Code. Anthropic later traced part of the drain to a caching bug fixed April 10.

Claude March 2026 promotion doubles usage limits outside peak hours

Anthropic ran a March 2026 promotion that doubled usage limits outside a six-hour peak window. The Register reports it ended on March 28, 2026. The Hacker News thread drew 255 points and 148 comments. Anthropic's help article is no longer online.

Who is affected:

Claude subscribers using Claude Code in March 2026.

What to do:

No action needed. The promotion has ended.

Anthropic's Claude Code terms bar routing Free, Pro or Max credentials through third-party products

Claude Code's legal and compliance page states that subscription OAuth is only for ordinary use of Claude Code and Anthropic's own apps, and developers building products must use API keys. The Hacker News thread drew 655 points and 792 comments.

Who is affected:

Builders of tools that sign users in with a Claude subscription.

Action needed:

Move third-party integrations to Claude Console API keys or a cloud provider.

Hacker News thread: Claude Code is being dumbed down?

A blog post argued that Claude Code 2.1.20 replaced file paths and search patterns with one-line summaries such as 'Read 3 files', hiding what the agent does. The thread drew 1,085 points and 701 comments.

Who is affected:

Claude Code users who review which files the agent reads.

What to do:

Use verbose mode if you need full file paths.

Claude Max logins stop working in third-party coding tools such as OpenCode

OpenCode users reported that Claude Max authentication stopped working on January 9, 2026, and an OpenCode maintainer said Anthropic had likely blocked it. The Hacker News thread drew 625 points and 513 comments.

Who is affected:

Developers using a Claude Pro or Max login inside third-party coding agents.

Action needed:

Use an API key from the Claude Console for third-party tools.

Anthropic confirms three infrastructure bugs degraded Claude responses in August and September 2025

Anthropic published a postmortem saying three separate infrastructure bugs, the first introduced on August 5, intermittently degraded Claude's response quality until early September. A load balancing change on August 29 increased affected traffic. The Hacker News thread drew 381 points and 116 comments.

Who is affected:

Users of Claude Sonnet 4, Opus 4 and 4.1, and Haiku 3.5 on the Claude API and in Claude Code between August 5 and mid-September 2025.

What to do:

No action needed. Anthropic says all three bugs are fixed.

Anthropic adds weekly usage limits to Claude subscriptions from August 28, 2025

Anthropic emailed subscribers that from August 28, 2025 an overall weekly limit and a separate Claude Opus 4 weekly limit would apply alongside the 5-hour limit, affecting less than 5% of users. It estimated 140 to 280 hours of Sonnet 4 and 15 to 35 hours of Opus 4 a week for most Max 5x users. The Hacker News thread drew 609 points and 705 comments.

Who is affected:

Claude Pro and Max subscribers who run Claude Code for long or parallel sessions.

Action needed:

Check your weekly usage and plan long Opus sessions around the weekly reset.

Claude Code users hit tighter usage limits without an announcement

TechCrunch reported that from July 14, 2025 heavy Claude Code users, many on the $200 Max plan, hit unexpectedly restrictive limits with no announced change. Anthropic confirmed issues. The Hacker News thread drew 407 points and 257 comments.