The newest Claude Opus model is Claude Opus 5.5, released on September 22, 2026. The line's usual gap is about 60 days and 32 have passed, so it is mid-cycle.
How long it has been since the last release against the line's usual gap and its recent ones — the rhythm, not a date.
Latest: Claude Opus 5.5 · Sep 22, 2026
Every Monday, only when something shipped: the AI models that came out, what each one changed, and which lines are due next.
We email you a link to confirm first · unsubscribe anytime · read past issues · about the newsletter · RSS for the Claude Opus line
Newest first — what each version changed versus the one before, in Anthropic's own numbers.
Anthropic's new Opus scores 7 points higher than Opus 5 on the all-round index and costs 20% less per token.
Compared with Claude Opus 5 · the model it replaces
Completes two in three hard command-line tasks
Terminal-Bench 4.0 — doing real work in a computer terminal
66.4%
▲ up from 52.3% · +14.1 points
Scores 7 points higher on the all-round intelligence index
Artificial Analysis Intelligence Index v4.3.2 — one score averaging 10 hard tests
58
▲ up from 51 · +7 points
Costs a fifth less than the model it replaces
Price per 1M tokens — what developers pay, input / output
$4 / $20
▲ down from $5 / $25 · 20% cheaper
Scores 74.6% on Terminal-Bench 2.1, up from 66.1% for Opus 4.7, at the same $5 / $25 price per 1M tokens, and its fast mode now costs a third of what fast mode cost for earlier models.
Compared with Claude Opus 4.7 · the model it replaces
Finishes more multi-step tasks at the command line
Terminal-Bench 2.1 — running real commands to complete a task
74.6%
▲ up from 66.1% · +8.5 points
Solves 7 in 10 hard real-world coding tasks
SWE-bench Pro — fixing real bugs in real software projects
69.2%
▲ up from 64.3% · +4.9 points
Costs the same to run as before
Price per 1M tokens — what developers pay, input / output
$5 / $25
same price as Opus 4.7
Anthropic's updated flagship: a clear jump on real-world coding and much sharper at reading charts than 4.6 — at the same price.
Compared with Claude Opus 4.6 · the model it replaces
Solves nearly 9 in 10 real-world coding tasks
SWE-bench Verified — fixing real bugs in real software projects
87.6%
▲ up from 80.8% · +7 points
Makes the right code edits 7 times in 10 inside an editor
CursorBench — editing code the way developers do in an IDE
70%
▲ up from 58% · +12 points (Cursor's own test, quoted by Anthropic)
Costs developers the same as before
Price per 1M tokens — what developers pay, input / output
$5 / $25
same price as Opus 4.6
Anthropic's Opus 4.6: a big jump on brand-new reasoning puzzles and the first Opus that can read 1M tokens of text at once — but its score on fixing real software bugs (SWE-bench Verified) is unchanged.
Compared with Claude Opus 4.5 · the model it replaces
Solves far more puzzle-style problems it has never seen before
ARC-AGI-2 — reasoning through brand-new problems, not memorized ones
68.8%
▲ up from 37.6% · +31 points
First Opus that can read 1M tokens of text at once
Context window — how much text it can hold in one go
1M tokens
✦ new in Opus 4.6 (in beta, Claude API only at launch) — earlier Opus models held less
Fixes real software bugs at the same rate as before
SWE-bench Verified — fixing real bugs in real software projects
80.8%
no change from Opus 4.5 — was 80.9%
Opus 4.5 fixes more real software bugs than Opus 4.1 (80.9% vs 74.5% on SWE-bench Verified) and costs two-thirds less, at $5 / $25 per 1M tokens instead of $15 / $75.
Compared with Claude Opus 4.1 · the model it replaces
Fixes about 8 in 10 real software bugs
SWE-bench Verified — fixing real bugs in real software projects
80.9%
▲ up from 74.5% · +6.4 points
Costs a third of what Opus 4.1 cost to run
Price per 1M tokens — what developers pay, input / output
$5 / $25
▲ down from $15 / $75 · two-thirds cheaper
The top model of the Claude 3 family, with a 200K-token context window, priced at $15 / $75 per 1M tokens.
Scores higher than Claude 1.3 on coding and math tests, reads up to 100K tokens per prompt, and opened as a public beta on claude.ai in the US and UK.
Compared with Claude 1.3 · comparison chosen by Anthropic
Writes correct Python code for 7 in 10 test problems
Codex HumanEval — short Python programming tasks
71.2%
▲ up from 56.0% for Claude 1.3 · +15.2 points
Solves almost 9 in 10 grade-school math problems
GSM8k — grade-school math word problems
88.0%
▲ up from 85.2% for Claude 1.3 · +2.8 points