According to VentureBeat, Anthropic's own testing shows Opus 5 scoring 43.3 percent on Frontier-Bench v0.1, an agentic terminal coding benchmark — more than double Opus 4.8's 18.7 percent and ahead of Fable 5's 33.7 percent, which the company says it achieved at a lower cost per task (E2).
But the same VentureBeat report notes Anthropic itself acknowledges gaps: Opus 5 "remains behind Mythos 5, a competing model, on cybersecurity tasks and biology research," and an OpenAI-family model still leads on one agentic coding benchmark (E3). On Anthropic's OSS-Fuzz vulnerability-discovery evaluation, Opus 5 found vulnerabilities at a 79.4 percent rate — close to Mythos 5's 80 percent — but only managed to develop working exploits in 4 challenges, versus 13 for Mythos 5 (E7). The pattern across these three data points is consistent: Opus 5 leads clearly in coding-agent scoring but trails specialized rivals in cybersecurity-adjacent tasks.
Per VentureBeat, Opus 5 launched with pricing unchanged from its predecessor, Opus 4.8: $5 per million input tokens and $25 per million output tokens (E1). The efficiency claim instead comes from token usage rather than list price. Harvey, the legal AI company, told VentureBeat that Opus 5 matched the performance of Opus 4.8's maximum-reasoning mode "while generating 26% fewer tokens on average," according to Niko Grupen, its head of applied research (E4). Since output tokens are billed at $25 per million, a customer-reported drop in token volume — at flat per-token pricing — is the basis for Anthropic's cost-efficiency narrative.
According to VentureBeat, Zapier CEO Wade Foster said Opus 5 topped his company's AutomationBench leaderboard "without spending more tokens than prior Claude models," completing a full customer-churn-prevention workflow from start to finish end-to-end. Foster said: "Previous models didn't pass; Opus 5 hit 100%" (E5).
TechCrunch reports that Anthropic paired the release with a new beta feature, Automatic Fallbacks, which lets users opt in to automatically route requests to a less powerful model whenever a prompt triggers the safety classifier (E13). Anthropic told TechCrunch it expects these safety classifiers to engage 85 percent less often for Opus 5 than for Fable 5, which the company frames as a reflection of the lighter oversight given to a less capable model (E14).
VentureBeat reports that Anthropic's automated behavioral audit found Opus 5 to be its "most aligned model to date," scoring 2.3 on overall misaligned behavior — lower than Opus 4.8, Sonnet 5, or Fable 5 — with the lowest rates of deceptive behavior and the least susceptibility to being tricked into misuse, according to the company (E6).
TechCrunch adds a data-handling detail: like its predecessor, Opus 5 is not subject to the 30-day data retention policy that covers Fable and Mythos, a policy that TechCrunch says had raised concerns among some privacy-conscious users (E12).
TechCrunch reports that Opus 5 is launching only two months after Opus 4.8, which became available on May 28 (E11). That cadence places the two releases in the same year, with the 2026-07-24 Opus 5 launch date recorded by both VentureBeat and TechCrunch coverage (E1, E11).
According to VentureBeat's citation of a February 2026 analysis by Contrary Research, Claude held roughly 40 percent of the enterprise large language model market by usage as of late 2025, and Claude Code alone had reached about $1 billion in annualized revenue (E8). VentureBeat also notes that Reuters reported in February that Anthropic was valued at roughly $380 billion in its latest funding round (E9).
Separately, VentureBeat reports that a U.S. judge gave final approval this week to Anthropic's $1.5 billion copyright settlement with book authors, closing a chapter of litigation over the company's early training data, according to Reuters (E10).
| Metric | Opus 5 | Opus 4.8 | Fable 5 | Mythos 5 | Source |
|---|---|---|---|---|---|
| Frontier-Bench v0.1 (coding) | 43.3% | 18.7% | 33.7% | — | E2 |
| OSS-Fuzz vulnerability discovery rate | 79.4% | — | — | 80% | E7 |
| OSS-Fuzz successful exploits | 4 challenges | — | — | 13 challenges | E7 |
| Misaligned behavior score | 2.3 (lowest) | higher | higher | — (Sonnet 5 also higher) | E6 |
| Token reduction vs. prior max-reasoning mode | 26% fewer (per Harvey) | baseline | — | — | E4 |
| Safety classifier trigger rate | 85% less than Fable 5 (expected) | — | baseline | — | E14 |
| Input/output pricing | $5 / $25 per million tokens | $5 / $25 per million tokens (unchanged) | — | — | E1 |
Taken together, the evidence points to a release built on flat pricing rather than a price cut: Opus 5 carries the same $5/$25 per-million-token rate as Opus 4.8 (E1), with the cost-efficiency case resting instead on Harvey's reported 26 percent token reduction (E4) and Zapier's reported 100 percent AutomationBench pass rate achieved without extra token spend (E5). At the same time, Anthropic's own disclosures show a mixed competitive picture — clear leads on Frontier-Bench v0.1 and alignment scoring (E2, E6), but acknowledged gaps against Mythos 5 in cybersecurity and biology tasks and in exploit-development capability (E3, E7). The two-month gap between Opus 4.8 and Opus 5 (E11) suggests Anthropic is iterating on the Opus line in rapid succession even as it reports a roughly 40 percent enterprise usage share and a $380 billion valuation (E8, E9), and closes out a $1.5 billion copyright settlement from its earlier training-data litigation (E10).
According to TechNews and China Times reports, Meta is preparing a $12 billion bond sale through a BlackRock-backed SPV for a Texas data center, with investors demanding yields above 7%—about 0.4 percentage points higher than Meta's October 2025 Hyperion deal. MoneyDJ reports Alphabet's Q2 free cash flow turned negative for the first time as capex surged, underscoring investor unease over AI spending.
According to a CNA report (carried also by CTS), Taiwan's King Slide (川湖) says its Houston plant will start mass production in the third quarter, while Chenbro (勤誠) has already opened a US NCT site in March and signed a Dallas plant deal in May, as both AI supply-chain suppliers post sharp first-half revenue gains.
According to CNA (中央社) and Technews (科技新報), new U.S. Section 301 tariffs took effect July 24, setting Taiwan's rate at 10% versus 12.5% for China and Japan. Taiwan Institute of Economic Research (台經院) president Chang Chien-yi called the outcome favorable, while analyst Liu Pei-chen said impact on Taiwan's chip sector is limited, with advanced nodes largely shielded and mature nodes more exposed to a pending overcapacity probe.