LLMs TigerLive LLM benchmarks

Claude Opus 4: benchmark scores, price and ranking

Claude Opus 4 from Anthropic ranks #319 of 743 LLMs on the LLMs Tiger aggregate benchmark score (48.9), based on 17 benchmark results captured up to 2026-10-13.

Organisation
Anthropic
Type
Frontier model
Open weights
No
Released
2025-05-22 (source: curated)
Cheapest API price
$15.00 in / $75.00 out per million tokens (via gmi, observed 2026-10-13T19:47Z)
Aggregate rank
#319 of 743 (score 48.9)

Relative to other models, Claude Opus 4 is strongest on SimpleBench (#10 of 61) and weakest on ARC-AGI-2 (#70 of 87).

Across the 16 benchmarks where at least ten models are scored, Claude Opus 4 is in the top 10% on 0 and in the top half on 12.

Its cheapest listed input price, $15.00 per million tokens, is lower than that of 2% of the 427 models with a listed price.

Claude Opus 4 benchmark results

Claude Opus 4 benchmark scores, ranked against all models measured on each benchmark
BenchmarkScoreRankSourceCaptured
SimpleBench58.8%#10 of 61Epoch AI Benchmarking Hub2026-10-13
Epoch ECI143#39 of 204Epoch AI Benchmarking Hub2026-10-13
SWE-bench Verified70.7%#18 of 73Epoch AI Benchmarking Hub2026-10-13
Cybench38.0%#4 of 16Epoch AI Benchmarking Hub2026-10-13
METR time horizon100 min#9 of 31Epoch AI Benchmarking Hub2026-10-13
EnigmaEval5.6%#7 of 24Epoch AI Benchmarking Hub2026-10-13
GSO-Bench6.9%#6 of 19Epoch AI Benchmarking Hub2026-10-13
DTBench81.6%#47 of 146Epoch AI Benchmarking Hub2026-10-13
LMCA37.4%#37 of 108Epoch AI Benchmarking Hub2026-10-13
WeirdML43.7%#44 of 109Epoch AI Benchmarking Hub2026-10-13
GPQA Diamond76.3%#121 of 275Epoch AI Benchmarking Hub2026-10-13
OTIS Mock AIME64.4%#88 of 189Epoch AI Benchmarking Hub2026-10-13
Fiction.liveBench61.1%#24 of 47Epoch AI Benchmarking Hub2026-10-13
Humanity's Last Exam10.7%#14 of 27Epoch AI Benchmarking Hub2026-10-13
DeepResearch Bench46.8%#4 of 7Epoch AI Benchmarking Hub2026-10-13
CritPt0.3%#75 of 115Epoch AI Benchmarking Hub2026-10-13
ARC-AGI-21.3%#70 of 87ARC Prize official leaderboard2026-10-13

About Claude Opus 4

Claude Opus 4 is a frontier model from Anthropic. It was released on 2025-05-22, according to the developer's announcement. The cheapest API listing we track charges $15.00 per million input tokens and $75.00 per million output tokens, through gmi. We hold 17 benchmark results for it across 4 areas, and none is in the top tenth of its board.

Where Claude Opus 4 stands

On our aggregate score, Claude Opus 4 is #319 of 743 ranked models, around the middle of the table. Its score is 48.9; the scale is centred on 50, so it sits 1.1 points below the typical model. That beats 57% of the ranked field. Among the 70 ranked Anthropic models it is #50. The next one up is claude-opus-4-5-20251101-high-32k at 48.9. The next one down is Claude 3.5 Haiku at 48.5.

Claude Opus 4, benchmark by benchmark

SimpleBench puts Claude Opus 4 at 58.8%, #10 of 61. That beats 84% of the models tested, but 23.1 points short of Claude Fable 5 (max) (81.9%). The results either side are claude-opus-4-1-20250805 (60.0%) and Kimi K2.7 Code (57.9%). It covers trick questions that test common sense. The board median is 41.4%, so Claude Opus 4 is 17.4 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

On Epoch ECI, Claude Opus 4 reaches 143. It ranks #39 of 204, ahead of 81% of the field, and 14 points behind Qwen 3.8 Max at 156. The results either side are Gemma 4 31B IT (143) and GPT-5.5 Instant (143). It covers Epoch Capabilities Index: Epoch's own aggregate of ~40 benchmarks. The board median is 127, so Claude Opus 4 is 15 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

Claude Opus 4 manages 70.7% on SWE-bench Verified, good for #18 of 73. Claude Opus 4.7 (max) leads at 83.5%, 12.8 points further up, and 75% of models sit below. The results either side are Kimi K2.5 (70.8%) and Claude 4.5 Sonnet (70.6%). It covers real GitHub issues resolved. The board median is 56.4%, so Claude Opus 4 is 14.3 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

Cybench puts Claude Opus 4 at 38.0%, #4 of 16. That beats 75% of the models tested, but 5.0 points short of grok-4-0709 (43.0%). The results either side are grok-4-1 (39.0%) and claude-sonnet-4-20250514 (35.0%). It covers cybersecurity capture-the-flag tasks. The board median is 17.5%, so Claude Opus 4 is 20.5 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

On METR time horizon, Claude Opus 4 reaches 100 min. It ranks #9 of 31, ahead of 71% of the field, and 944 minutes behind Claude Mythos Preview (Early) at 1,045 min. The results either side are claude-opus-4-1-20250805 (100 min) and claude-sonnet-4-20250514_16K (75 min). It covers length of task (minutes) done at 50% success. The board median is 23 min, so Claude Opus 4 is 77 minutes above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

Claude Opus 4 manages 5.6% on EnigmaEval, good for #7 of 24. Gemini 3 Pro Preview leads at 18.2%, 12.7 points further up, and 71% of models sit below. The results either side are Claude Opus 4.1 (unknown thinking) (7.2%) and Gemini 2.5 Pro Preview (Jun 2025) (5.6%). It covers multi-step puzzle hunts. The board median is 2.9%, so Claude Opus 4 is 2.7 points above it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

Claude Opus 4 lands below the middle on Fiction.liveBench: 61.1%, #24 of 47, with 49% of the field scoring higher. grok-4-0709 leads at 94.4%; the closest result above is qwen3-8b at 62.1%. It covers long-context story comprehension (16k). The median score on this board is also 61.1%. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

Claude Opus 4 lands below the middle on Humanity's Last Exam: 10.7%, #14 of 27, with 48% of the field scoring higher. Muse Spark leads at 40.6%; the closest result above is gemini-2.5-flash-preview-05-20 at 11.0%. It covers expert-level questions across domains. The median score on this board is also 10.7%. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

Claude Opus 4 lands below the middle on CritPt: 0.3%, #75 of 115, with 64% of the field scoring higher. GPT-5.6 Sol (max) leads at 32.3%; the closest result above is MiniMax-M2.7 at 0.6%. It covers unpublished research-level physics problems. The board median is 2.9%, so Claude Opus 4 is 2.6 points below it. Source: Epoch AI Benchmarking Hub, captured 2026-10-13.

ARC-AGI-2 is a weak spot. Claude Opus 4 scores 1.3%, #70 of 87, behind 79% of the models measured. The leader, GPT-6 Astra (max), reaches 95.0%. It covers abstract reasoning, ARC Prize public eval set. The board median is 22.8%, so Claude Opus 4 is 21.5 points below it. Source: ARC Prize official leaderboard, captured 2026-10-13.

The full table above also lists the other 7 results, including boards with fewer than ten models, where a rank says little.

Strengths and weaknesses

Software engineering: solid, ahead of 72% of the field on average across 2 benchmarks. Its best board here is SWE-bench Verified (#18 of 73) and its weakest is GSO-Bench (#6 of 19). Reasoning, maths and science: solid, ahead of 52% of the field on average across 7 benchmarks. Its best board here is SimpleBench (#10 of 61) and its weakest is ARC-AGI-2 (#70 of 87). General knowledge and chat: solid, ahead of 67% of the field on average across 6 benchmarks. Its best board here is Epoch ECI (#39 of 204) and its weakest is Fiction.liveBench (#24 of 47).

Price and value

The median input price across the 427 priced models is $0.500 per million tokens, so Claude Opus 4 costs more than the typical listing. Models that score within 1.5 points or better and cost less: Qwen2.5-Coder-7B-Instruct ($0.010, 49.5), DeepSeek V4.1 Flash (high) ($0.010, 48.9), Qwen3.7 Flash ($0.030, 50.5). Similar scores cost more on GPT-4 (0125) ($30.00) and gpt-4.5-preview ($75.00).

Release history

The Anthropic model released before it was Claude 4 Sonnet on 2025-05-22, which scores 54.5. The next one was claude-opus-4-20250514 on 2025-05-22, at 52.5.

Reading these numbers

The 17 scores on this page come from 2 sources: Epoch AI Benchmarking Hub (16), ARC Prize official leaderboard (1). 1 of them come from an official leaderboard and 16 from datasets or third-party runs. The leaderboard also lists 3 other reasoning-effort settings of this model, which redirect here. This page shows the highest.

Claude Opus 4 is also listed at 3 other reasoning-effort settings; this page shows the highest-effort row, which is the one the leaderboard keeps by default.

Models ranked nearby

More from Anthropic

Compare it on the live leaderboard → · How the aggregate score works