Claude Opus 4.8 — AI Model Review & Pricing
Claude Opus 4.8, released May 2026, is Anthropic's latest general-access Opus-tier model. ranked #2 on LMRank, it leads on coding and reasoning benchmarks with a 1M token context window. Available via Anthropic API, claude.ai, AWS Bedrock, GCP Vertex AI, and Microsoft Foundry.
Specifications
| Attribute | Value |
|---|---|
| Lab | Anthropic |
| Tags | Intelligent Reasoning Agentic Multimodal |
| Release Date | 2026-05 |
| Context Window | 1,000,000 tokens |
| Input Price / 1M | $15.00 |
| Output Price / 1M | $75.00 |
| Input Modalities | Text |
| Output Modalities | Text |
Strengths
- Top-tier coding and reasoning benchmarks
- 1M token context window
- Strong agentic workflow support
- Wide cloud provider availability
Weaknesses
- Higher latency (58.7 tokens/s output)
- Premium pricing vs. median models
- No extended thinking budgets or top-p sampling
- Verbose output above median token usage
Best For
- Complex code generation and codebase migrations
- Agentic orchestration and long-form analysis
- Research synthesis with large contexts
- End-to-end project planning
In Depth: Claude Opus 4.8
Benchmark Performance
Claude Opus 4.8 and ranks #2 overall on LMRank. At launch, it claimed the #1 spot on Artificial Analysis's Intelligence Index (max effort), outperforming GPT-5.5 and prior Opus models across coding and reasoning suites.
On SWE-bench Verified, Opus 4.8 scores 88.6%, and on SWE-bench Pro it reaches 69.2%, beating GPT-5.5 by +10.6 points. It also achieves 84.4% on SWE-bench Multilingual, climbs +6.8 on Terminal-Bench Hard over Opus 4.7, and gains +5.9 on τ²-Bench Telecom. On AA-Omniscience it ranks #2 with 27.4, behind Gemini 3.1 Pro (32.9).
Pricing & Value
Official pricing is $5 per million input tokens and $25 per million output tokens (standard), with $10/$50 for fast mode (research preview). Prompt caching delivers a 90% discount on cache hits.
At standard rates, Opus 4.8 costs $15 per million input and $75 per million output on some resellers confirm official Anthropic pricing. Compared to GPT-5.5 and Gemini 3.1 Pro, Opus 4.8 delivers leading coding benchmark scores but at a higher per-token cost. The 1M context window adds value for long-document tasks, while the 90% cache discount reduces ongoing expenses for repeated queries.
Who Should Use This
Senior software engineers handling large-scale migrations. Research analysts synthesizing lengthy documents. Teams building agentic pipelines. Not ideal for latency-sensitive applications.
- Senior engineers: excels at codebase-scale migrations; tradeoff is slower generation speed
- Research analysts: powerful for long-form synthesis with 1M context; verbosity may add overhead
- Agentic workflow builders: strong orchestration capabilities; slightly reduced prompt injection robustness vs. Opus 4.7
- Anti-persona: teams needing low latency or strict budgets should evaluate cheaper or faster alternatives
Release & Version History
Claude Opus 4.8 launched May 28, 2026 as the successor to Opus 4.7, Anthropic's previous general-access flagship. It quickly became the #1 model on Artificial Analysis's Intelligence Index at launch.
Anthropic released Opus 4.8 on May 28, 2026 (OpenRouter shows May 27 due to timezone), succeeding Opus 4.7. It is distinct from the more capable internal Claude Mythos Preview. Opus 4.8 delivers incremental gains across coding and reasoning benchmarks, with a wider context window and expanded cloud availability. No successor has been announced as of this writing.