Reading
Opus 5.5 Beats Fable for Less. Is Claude Now the Best AI Deal?
Anthropic cut Opus prices and claims Fable-level results. Here's what changed, who benefits, and why the benchmark bragging needs a reality check.
Prices last read 2026-09-28
Short answer
For API developers, probably yes. Opus 5.5 is Anthropic's cheapest Opus yet, and Anthropic's own numbers put it ahead of Fable 5.1 on coding and agent tests. But those are the company's own figures, so wait for independent tests before calling it the best deal.Anthropic released Claude Opus 5.5 on September 22, 2026. API pricing is $4 per million input tokens and $20 per million output tokens, 20% below Opus 5. Cache reads dropped to $0.20 per million, a 60% cut. Anthropic says the model generates output more than 30% faster than Opus 5. Anthropic says Opus 5.5 is available on the Pro, Max, Team and seat-based Enterprise plans, and it raised five-hour usage limits on those plans.Source: Introducing Claude Opus 5.5 — Anthropic · 2026-09-28; Anthropic upgrades Claude with new Opus 5.5 model — 9to5Mac · 2026-09-28
What actually changed
Price is the headline. On Anthropic's launch page, cache writes cost $5 per million tokens and a new fast mode costs $8 input and $40 output. 9to5Mac quotes Anthropic's claim that Opus 5.5 will cost 40% less than Opus 5 on typical workloads. That figure is higher than the 20% list-price cut because of the cheaper cache reads. The model ID is claude-opus-5-5, and it runs on Anthropic's API, AWS, Google Cloud and Microsoft Azure.
Source: Introducing Claude Opus 5.5 — Anthropic · 2026-09-28; Anthropic upgrades Claude with new Opus 5.5 model — 9to5Mac · 2026-09-28
On benchmarks, Anthropic's Terminal-Bench 4.0 table shows Opus 5.5 at 66.4%, GPT-6 Astra at 57.9%, Fable 5.1 at 55.8% and Opus 5 at 52.3%. VentureBeat's rundown also lists wins over Fable 5.1 on FrontierCode v1.1 (54.4% vs 50.3%), CursorBench 4.0 (57.8% vs 51.8%) and GDPval-AA v2.1 (1846 vs 1735 Elo). MacRumors reports that Anthropic calls it comparable to Fable 5.1, not better, in biology and cybersecurity.
Source: Introducing Claude Opus 5.5 — Anthropic · 2026-09-28; Anthropic releases Claude Opus 5.5, beating Fable 5.1 on key agentic benchmarks — VentureBeat · 2026-09-28; Anthropic Launches Claude Opus 5.5 With Fable-Level Performance at a Lower Price — MacRumors · 2026-09-28
The fine print nobody puts in the headline
Anthropic's own page lists some limits. Cybersecurity tasks can be re-routed to Opus 4.8. When safeguards step in on some biology and frontier-AI development tasks, Claude Opus 5 completes them instead. Thinking mode can't be turned off. VentureBeat notes that moving from Opus 5 requires API changes. VentureBeat also says there was no same-harness head-to-head against GPT-6 Astra. Anthropic's footnote gives a standard error of ±2.6 points for the Opus 5.5 Terminal-Bench score.
Source: Introducing Claude Opus 5.5 — Anthropic · 2026-09-28; Anthropic releases Claude Opus 5.5, beating Fable 5.1 on key agentic benchmarks — VentureBeat · 2026-09-28
Our take: the price cut is real, the benchmark lead is marketing for now
Cheaper and faster at the same time is the rare AI launch that helps your wallet more than the press release. The 60% cut on cache reads is the underrated part. If you run agents or long coding sessions against the same large context, that is where the savings pile up. The benchmark lead over Fable 5.1 and GPT-6 Astra is a different story. These are the vendor's own tables, and the vendor picked the tests. A win of roughly 9 to 11 points on Terminal-Bench is big if it holds up. Treat it as marketing until independent tests confirm it. Still, this puts real pressure on ChatGPT and Gemini to answer on price rather than just on leaderboard screenshots.
Who is affected and what to do
Your move, by user type
- API developers on Opus 5: plan the migration. VentureBeat says API changes are needed, so test in staging before switching the model ID.
- Heavy prompt-caching users: re-run your cost estimates. Cache reads fell 60%, according to Anthropic.
- Pro, Max, Team and seat-based Enterprise subscribers: Anthropic raised five-hour usage limits and gave subscribers a rate-limit reset they can save and use later. MacRumors reports it can be used through October 22.
- Security researchers and biology teams: expect some requests to be handled by fallback models (Opus 4.8 or Opus 5), per Anthropic.
- Deciding between Claude, ChatGPT and Gemini: run your own real tasks on each before switching, instead of relying on the launch charts.
Source: Introducing Claude Opus 5.5 — Anthropic · 2026-09-28; Anthropic Launches Claude Opus 5.5 — MacRumors · 2026-09-28; Anthropic releases Claude Opus 5.5 — VentureBeat · 2026-09-28
More models are coming. TechCrunch reports that Sonnet 5.5 and Haiku 5.5 will arrive in the coming weeks, and that outside evaluators including METR tested Opus 5.5 before release.
Source: Anthropic releases Opus 5.5 with lower prices and Fable-level performance — TechCrunch · 2026-09-28
the Company may add or remove features, increase or decrease capacity limits, offer new services, or discontinue certain services
Anthropic Consumer Terms of Service (effective October 8, 2025)Source: Consumer Terms of Service — Anthropic · 2026-09-28
Why this clause matters: under Anthropic's own terms, the higher usage limits and the model lineup that come with this launch can change later. Enjoy the higher limits, but don't build a workflow that assumes they are permanent.
Source: Consumer Terms of Service — Anthropic · 2026-09-28
Optional: compare Claude subscription listings from sellers (a seller listing, not official Anthropic access or an endorsement). That link carries our partner code Sk52g, which is how a purchase is attributed to this site. Promotion code UPD3S.
Sources (read 2026-09-28)
- Introducing Claude Opus 5.5 — Anthropic
- Anthropic releases Opus 5.5 with lower prices and Fable-level performance — TechCrunch
- Anthropic releases Claude Opus 5.5, beating Fable 5.1 on key agentic benchmarks — VentureBeat
- Anthropic Launches Claude Opus 5.5 With Fable-Level Performance at a Lower Price — MacRumors
- Anthropic upgrades Claude with new Opus 5.5 model — 9to5Mac
- Consumer Terms of Service — Anthropic