Compare Start here

AnthropicLanguage modelSuperseded

Claude Opus 4.8

Refined Opus 4 generation; later the safety fallback model for Fable 5.

Compare head-to-head Race it in 3D

Overall Intelligence

#18

of 56
7.5 pts behind #1

0

races led

44

weeks at #1

81.2

score

Released: May 14, 2026; 5 mo ago
Context: 1M; 64K max output
Price: $5 / $25; per 1M in/out · $10.00 blended
Speed: 66 tok/s; 5s to first token
Modalities: text · image; outputs text
Weights: Closed; API and apps

The 30-second version

Refined Opus 4 generation; later the safety fallback model for Fable 5.

Key takeaways

1Refined Opus 4 generation; later the safety fallback model for Fable 5.

Where it ranks

Current snapshot · illustrative scores · click any race to watch it

Opus 4.8Best on each axis

Overall

18Overall+7.5 pts

Coding

16Coding+8.2 pts 11Frontend & UI+41 Elo 10Backend & APIs+6.2 pts 17Distributed Systems+10.6 pts 13Full-Stack Apps+8.5 pts 16Debugging & Repair+9.8 pts 13Agentic Coding+9.0 pts 16Code Review & Security+8.0 pts 18Data Science & SQL+7.6 pts 8Mobile Apps+5.0 pts 11Terminal & DevOps+10.0 pts 20Competitive Programming+287

Hard Reasoning

18Hard Reasoning+8.7 pts 18Frontier Questions+15.7 pts 20Abstract & Novel Reasoning+14.8 pts 19Logic & Puzzles+7.4 pts 12Planning & Strategy+7.1 pts

Math

19Math+10.4 pts 21Competition Math+8.2 pts 19Research Math+14.5 pts 19Applied & Quantitative+8.0 pts

Science

19Science+8.9 pts 18Graduate Science QA+5.6 pts 13Biology & Medicine+6.6 pts 22Physics & Chemistry+9.5 pts 18Research Assistance+7.8 pts

Writing & Language

12Writing & Language+6.4 pts 10Creative Writing+38 Elo 9Technical Writing+5.0 pts 14Editing & Style+6.0 pts 12Multilingual+4.5 pts 16Instruction Following+6.2 pts 14Conversation & EQ+39 Elo

Multimodal

27Multimodal+21.3 pts 23Image Understanding+10.8 pts 22Documents & Charts+10.7 pts 24Video Understanding+21.7 pts 33Speech & Audio+32.4 pts

Tool Use & Agents

13Tool Use & Agents+7.9 pts 11Tool Use+5.7 pts 13Computer Use+13.0 pts 16Web Research+10.7 pts 11Long-Horizon+18.1 hrs

Long Context

21Long Context+10.8 pts 221M Recall+9.5 pts 21Long Reasoning+9.0 pts

Speed & Cost

44Speed & Cost+48.1 pts 41Output Speed+457 tok/s 37Latency+4.70s 40Cost-Efficiency+53.5 pts

Trust & Factuality

13Trust & Factuality+7.7 pts 11Factuality+6.1 pts 13Honesty & Calibration+6.9 pts

Overall Intelligence over time

#1 frontierOpus 4.8Spark OpenKimi K3

Update history

The distilled version first; open any release for the deeper breakdown.

The Claude Opus family

Opus 4May 2025 Opus 4.1August 2025 Opus 4.5November 2025 Opus 4.6February 2026 Opus 4.8May 2026 Opus 5July 2026 Opus 5.5September 2026

60-second briefing

Opus 4.8, explained

AI video summary · placeholder render from the auto-written script

From Anthropic's own guide, distilled

Prompt engineering and Claude best practices

Be explicit about the outcome you want; newer Claude models follow instructions literally rather than guessing at hidden intent.
Explain why a rule exists. Context about the goal lets the model generalise the rule to cases you did not list.
Wrap documents, examples and data in XML-style tags so instructions and material never blur together.
Put long documents at the top of the prompt and your question at the end; recall and grounding both improve.
For agentic tasks, state when to act versus when to ask, and ask for a short plan before large edits.

All official guides, distilled →

About Anthropic

Safety-focused lab building the Claude family of models.

Optimising for

Reliable long-horizon agentic work (especially coding and computer use), honest and calibrated answers, and strong writing, with safety research shipped alongside capability.

San Francisco, US · founded 2021

Closest rivals

Opus 4.8 is 18th in overall intelligence; leader is Claude Opus 5.5.