# Anthropic Launches Claude Opus 5, Tops AI Benchmark Index at Half the Cost of Fable 5 > Source: > Published: 2026-07-26 09:22:09.518856+00:00 # Anthropic Launches Claude Opus 5, Tops AI Benchmark Index at Half the Cost of Fable 5 - Claude Opus 5 ranks first on the Artificial Analysis Intelligence Index with 61 points, ahead of Fable 5 (60) and GPT-5.6 Sol (59) [[1]](https://the-decoder.com/anthropics-claude-opus-5-costs-well-below-fable-5-while-matching-or-beating-it-across-most-benchmarks/) - Token pricing stays at $5 input / $25 output per million tokens — the same as predecessor Opus 4.8 — with average task costs roughly half those of Fable 5 [[2]](https://www.anthropic.com/news/claude-opus-5) - The model more than doubles Opus 4.8's Frontier-Bench score and scores roughly 3x higher than the next-best model on ARC-AGI 3 [[3]](https://thenextweb.com/news/anthropic-claude-opus-5-launch-frontier-bench-coding) - Anthropic describes Opus 5 as its most aligned model to date, with the lowest rate of deceptive behavior in its automated safety audit [[3]](https://thenextweb.com/news/anthropic-claude-opus-5-launch-frontier-bench-coding) - Opus 5 is Anthropic's fourth model shipped in eight weeks, following Fable 5, Sonnet 5, and Opus 4.8 [[4]](https://techcrunch.com/2026/07/24/anthropic-launches-opus-5/) Anthropic on July 24 released Claude Opus 5, a frontier AI model that tops the Artificial Analysis Intelligence Index with 61 points while costing roughly half as much per task as the company's own Fable 5. The model is priced at $5 per million input tokens and $25 per million output tokens — identical to its predecessor, Opus 4.8 [[1]](https://the-decoder.com/anthropics-claude-opus-5-costs-well-below-fable-5-while-matching-or-beating-it-across-most-benchmarks/) [3]. Opus 5 is Anthropic's fourth model launch in under two months, following Fable 5 in early June, Sonnet 5 at the end of June, and the Opus 4.8 update before that. The rapid cadence underscores an intensifying race at the top of the AI benchmark stack, where Opus 5 (61 points), Fable 5 (60), and OpenAI's GPT-5.6 Sol (59) are separated by just two points on the Intelligence Index [[1]](https://the-decoder.com/anthropics-claude-opus-5-costs-well-below-fable-5-while-matching-or-beating-it-across-most-benchmarks/) [4]. The model is available immediately on the Claude API, Claude.ai, Claude Code, Claude Cowork, Amazon Bedrock, Google Cloud, and Microsoft Foundry. It becomes the default model on Claude Max and the strongest model available on Claude Pro [3]. ## Benchmark Performance On coding evaluations, Opus 5 more than doubles its predecessor's Frontier-Bench score and matches or exceeds Fable 5 across most tasks. On the Artificial Analysis Coding Agent Index at extended-high effort, Opus 5 ties GPT-5.6 Sol at 67 points. On Terminal-Bench v2.1 at maximum effort, it scored 89%, again matching Sol [1]. Knowledge-work benchmarks show a wider gap. On Artificial Analysis's AA-Briefcase evaluation at maximum effort, Opus 5 earned a 1,720 Elo rating — 146 points ahead of Fable 5's 1,574. Its Analytical Quality Elo reached 2,016, roughly 300 points above Fable 5 [1]. On ARC-AGI 3, a novel problem-solving benchmark, the model scored roughly three times higher than the next-best competitor. On Zapier's AutomationBench, it achieved a 100% pass rate on a full churn-prevention sequence [3]. The model trails in a few areas. On the Epoch AI Overall Score, Opus 5 recorded 159, just below Fable 5's 161. Its hallucination rate stands at 50% — seven percentage points higher than Opus 4.8. It also remains behind a model called Mythos 5 on cybersecurity tasks [[1]](https://the-decoder.com/anthropics-claude-opus-5-costs-well-below-fable-5-while-matching-or-beating-it-across-most-benchmarks/) [3]. ## Pricing and Cost Advantage While Opus 5's per-token pricing is flat versus its predecessor, the cost advantage over Fable 5 is significant on a per-task basis. The average Intelligence Index task costs $2.03 with Opus 5 versus $2.75 with Fable 5 [1]. On AA-Briefcase tasks at the 'high' reasoning tier, Opus 5 costs $10.41 per task — 54% less than Fable 5's $22.30. Even at 'max' effort, which takes over 36 minutes per task, the cost is $17.79, still below Fable 5 [1]. A new fast mode runs 2.5x faster at double the base price, targeting latency-sensitive production workloads [3]. ## What Partners Are Saying Early integration partners highlighted the model's debugging and agentic coding strengths. Cognition CEO Scott Wu said Opus 5 'approaches Fable-level performance at half the cost' with 'particular strength on difficult debugging and root-cause analysis tasks' [3]. Cursor co-founder Sualeh Asif described Opus 5 as delivering 'near Fable 5 intelligence at Opus speed and cost.' Lovable co-founder Fabian Hedin reported 22% improvement on the hardest agentic coding tasks 'with far less variance run to run' [3]. ## Safety and Alignment Anthropic described Opus 5 as its most aligned model to date, scoring 2.3 on misaligned behavior — the lowest of its recent models. Safety safeguards intervene approximately 85% less frequently than with Fable 5 [3]. The company intentionally excluded dedicated cybersecurity training from Opus 5 but noted improvements on those tasks through general capability gains. A new Cyber Verification Program provides restricted-access users with additional capabilities [3]. ## Competitive Landscape The launch extends Anthropic's sprint to dominate the top of AI benchmark leaderboards during the summer of 2026. With four models shipped since late May, the company has blanketed the market from the budget tier (Sonnet 5) through the premium tier (Fable 5), with Opus 5 now occupying the cost-efficient frontier sweet spot [4]. The gap at the top of the Intelligence Index remains narrow: one point separates Opus 5 from Fable 5, and two points from GPT-5.6 Sol. Kimi K3 sits at 57 points. How long Opus 5 holds the top position depends on OpenAI's and Google's response cadence [1]. ## Companies mentioned ## Further sources The stories that matter, in one email. Free — unsubscribe anytime.