Anthropic
Anthropic ships Claude Opus 5.5, 90 minutes ahead of OpenAI's GPT-6 Sol and Luna
$4/$20 per million tokens, Fable-level performance claimed, three labs move on price within two days.
The answer
Anthropic released Claude Opus 5.5 on 22 September 2026 at $4/$20 per million tokens, 20% below Opus 5.
Anthropic released Claude Opus 5.5 on 22 September 2026, on the Claude Platform, AWS, Google Cloud and Microsoft Azure, callable as claude-opus-5-5. It replaces Opus 5, launched 24 July.
It performs at the level of Claude Fable 5.1 on most work and costs 40% less to run than Opus 5.
Pricing
| Per million tokens | Opus 5.5 | Opus 5 |
|---|---|---|
| Input | $4 | $5 |
| Output | $20 | $25 |
| Cache reads | $0.20 | $0.50 |
| Cache writes | $5 | $6.25 |
A fast mode runs at up to 2.5x speed for $8 input / $40 output per million tokens. Anthropic says output generation overall is more than 30% faster than Opus 5.
Benchmarks and safety
On Anthropic's figures: 66.4% on Terminal-Bench 4.0 (Opus 5 52.3%, GPT-6 Astra 57.9%); 54.4% on FrontierCode v1.1 (Fable 5.1 50.3%); 57.8% on CursorBench 4.0; 1846 Elo on GDPval-AA v2.1 (Fable 5.1 1735); 81.8% on OSWorld 2.0.
Opus 5.5 was tested pre-release by external evaluators Frontier Design and METR. Anthropic says it is the strongest model it has tested on its automated behavioural audit and ties Fable 5.1 for the lowest prompt-injection success rate on a Gray Swan evaluation. It ships with cybersecurity and biology safeguards similar to Fable 5.1's. Sonnet 5.5 and Haiku 5.5 follow in the coming weeks, Anthropic said.
Timing and competitors
The release is Anthropic's first since CEO Dario Amodei's 12 September essay 'We Must Pace the Frontier', which called for the industry to deliberately slow capability gains so safety work can keep pace.
OpenAI released GPT-6 Sol and GPT-6 Luna about 90 minutes after Opus 5.5, per TechCrunch, priced at roughly half the GPT-5.6 generation ($2/$10 for Sol, $0.10/$0.50 for Luna). xAI had released Grok 4.7 the day before, 21 September, at unchanged $2/$6 per million token pricing. Three labs moved on price or capability inside 48 hours.
Anthropic frames Opus 5.5 as its first release since Amodei's essay, arguing pacing is about safety practices keeping up with capability, not a freeze on shipping. The launch page does not link the underlying reports from Frontier Design, METR or Gray Swan; the benchmark and safety figures above are Anthropic's own account of that testing.
Frequently asked questions
What does Claude Opus 5.5 cost?
How does it compare on benchmarks?
Did OpenAI really launch minutes later?
What comes next from Anthropic?
Sources
- Claude Opus 5.5 — Anthropic, 22 September 2026
- Anthropic releases Opus 5.5 with lower prices and Fable-level performance — TechCrunch, 22 September 2026
- OpenAI launches GPT-6 Sol and Luna, boasting lower cost and fewer mistakes — TechCrunch, 22 September 2026
- We Must Pace the Frontier — Dario Amodei / Anthropic, 12 September 2026
- Introducing Grok 4.7 — xAI, 21 September 2026