Claude Opus 4.8 vs MiniMax M3: pricing, Quality, Value, and benchmarks

Buyer access

Pricing, app access, and Conversation Value

• Humanity's Last Exam

  Normalized quality input

  49.8%

  Anthropic Claude Opus 4.8 release page | Vendor-reported Anthropic Opus 4.8 HLE no-tools score. Do not replace with tools-enabled or adaptive-effort HLE variants.

• SWE-Bench Pro

  Software engineering task resolution

  69.2%

  Anthropic Claude Opus 4.8 release page | Vendor-reported Anthropic Opus 4.8 SWE-Bench Pro score. Do not substitute SWE-Bench Verified.

• Humanity's Last Exam

  Normalized quality input

  37.1%

  Artificial Analysis - Humanity's Last Exam evaluation | Third-party benchmark evaluation page used only after the official HLE leaderboard sources fail to yield a usable result.

Claude Opus 4.8 is Anthropic's newest Opus model, strongest for coding, agentic tasks, and complex professional work where the vendor-reported benchmark evidence applies.

Released May 28, 2026, Claude Opus 4.8 is Anthropic's current flagship and its most capable publicly available model. It is best suited to complex agentic coding, legal and financial document analysis, deep multi-step reasoning, and long-running autonomous tasks. The meaningful upgrades over 4.7 are a dramatic improvement in mathematical reasoning, meaningfully better honesty (it flags its own mistakes rather than quietly moving on), and efficiency gains that mean it uses around 35% fewer output tokens to do the same work — so you actually get a little more for your money despite the unchanged rate card. The honest caveats: it is expensive at $25 per million output tokens, which adds up fast on any high-volume or long-session workflow. On claude.ai, users now have control over the amount of effort Claude puts into a task, but Pro plan rate limits are real and noticeable if you push it hard — heavy users will hit the ceiling. It is also slower than average at inference speed, so it thinks longer before responding. For chat, summarisation, and general Q&A, Sonnet 4.6 covers 90%+ of workloads at 40% lower per-token cost — most buyers do not need Opus for everyday tasks. AnthropicFinout Bottom line: Opus 4.8 is genuinely the best model for serious, sustained, complex work. It is overkill and quietly costly for anything routine — and if you hit the rate limits on a Pro plan, the frustration will feel disproportionate to what you are paying.

This model is still under editorial review. We will publish a verdict as soon as we have completed our review of the AI model.

This model is still under editorial review. We will publish a verdict as soon as we have completed our review of the AI model.