whatmodel.sh / Claude Opus 4.8

Claude Opus 4.8

+ contribute

Claude Opus 4.8 is a closed-weights language model from Anthropic, released May 2026.

78.1±0.4#19 of 719 LLMsfrom 62 benchmark results across 17 sources

$ how it compares

The LLMs ranked either side of Claude Opus 4.8, out of 828 with a score, then more from Anthropic.

#	model	score	where	RAM	released
17	GPT-5.6 Terra OpenAI	79±1	api $4.50/M		2026-07
18	Qwen3.8 Max (2.4T-A95B) Alibaba	78±1	api $3.00/M +local	~380GB	2026-08
19	Claude Opus 4.8 Anthropic	78±0	api $10.00/M		2026-05
20	GPT-5.4 OpenAI	78±1	api $5.63/M		2026-03
21	DeepSeek V4.1 Flash DeepSeek	78±1	api $0.26/M +local	~537GB	2026-09
	more from Anthropic
2	Claude Opus 5.5 Anthropic	87±1	api $8.00/M		2026-09
4	Claude Fable 5.1 Anthropic	85±1	api $20.00/M		2026-09
5	Claude Sonnet 5.5 Anthropic	84±1	api $4.00/M		2026-09

see all LLMs · find the ones that fit your machine

$ what it's good at

agent tasks▁▁▃▆█▆▅▅▅▄▃▁ #19 of 300 · 28% of its score · 19 tests

Beats GPT-5.5, Hy4 preview (770B-A49B), GPT-5.6 Terra here, though they rank higher overall. Trails GLM-5.3, Claude Haiku 5.5, DeepSeek V4.1 Flash here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
SWE-bench Pro (split unspecified) lab	69.2% ✎	▄█·▄·▄██···▄ beats 3/9	very hard · avg 7.6%	▇▇ 4.1%	sharp · at its level
Arena Agent (October 2026 aggregate IPS)	0.0664 IPS ✎	▁▁▁▄▂▃▇█▃▂▁▃ beats 40/49	avg -0.312 IPS	▇ 3.5%	sharp
Terminal-Bench 4.0	23.6% ✎	█▅▅▅▅·▅▃▃▅▅▅ beats 9/20	very hard · avg 0.4%	▇ 2.2%	sharp · at its level
SWE-bench Multilingual lab	84.4% ✎	▃··▃·▅▃▃█··▃ beats 3/9	hard · avg 31%	▇ 2.0%	typical
SWE-bench Verified lab	88.6% ✎	▁▁▁▁▁▁▂▄█▆▃▁ beats 128/131	hard · avg 28%	▇ 1.9%	sharp
Chartography with tools lab	75.0% ✎	▃········▃██ beats 1/7	very hard · avg 4.7%	▇ 1.8%	sharp · at its level
τ³-bench banking_knowledge (v1.0.1 grading)	39.7% ✎	▃▃▂▃·▆▃▂█▅▅▂ beats 17/25	very hard · avg 4.1%	▇ 1.4%	at its level
E2E-SWE (build a working codebase from scratch, 186 tasks) pass@1	51.9% ✎	▃··▅··▃█▃▅▃▅ beats 9/12	very hard · avg 3.1%	▇ 1.4%	sharp · at its level

11 more tests

test	result	vs the others in its run	difficulty	counts	why
Terminal-Bench 2.1 lab	82.7% ✎	▁·▁·▁▁▄▃▄█▃▇ beats 41/56	hard · avg 17%	▇ 1.4%	typical
OSWorld-Verified lab	83.4% ✎	█▄·▂▆▂▄▂▆▄▆█ beats 22/25	hard · avg 29%	▇ 1.2%	typical
Humanity's Last Exam with tools lab	57.9% ✎	▁······▁·▄·█ beats 4/10	very hard · avg 14%	▇ 1.1%	at its level
DeepSWE v1.1 (resolved) lab	59.0% ✎	█·····████·█ beats 0/5	hard · avg 18%	▇ 1.0%	at its level
Toolathlon (Tool Decathlon) lab	59.9% ✎	█···█··█·███ beats 4/5	hard · avg 16%	▇ 1.0%	at its level
MCP-Atlas (public set)	77.8% ✎	█··█·██·████ beats 7/7	hard · avg 29%	▇ 0.9%	typical
BrowseComp with context management lab	84.3% ✎	▅·▅▃·█▃·█▅·▃ beats 10/14	medium · avg 44%	▇ 0.8%	typical
Legal Research Bench (agentic, all-pass)	42.9% ✎	▄█··▄▄▄██·▄█ beats 12/12	very hard · avg 5.8%	▇ 0.7%	at its level
RWSearch	77.3% ✎	████··█····█ beats 1/5	medium · avg 44%	▇ 0.6%	typical
KlinikeBench strict pass@1 (333 interactive clinical encounters, Terminus-2)	16.8% ✎	▁▁·▁▁▁▁▅▃█▅█ beats 17/27	very hard · avg 2.4%	▇ 0.6%	typical
Toolathlon-Verified lab	79.9% ✎	▃·····█···▃▃ beats 4/5	hard · avg 35%	▇ 0.5%	blunt · at its level

coding▁▁▁▃▆██▇▆▄▃▁ #34 of 458 · 15% of its score · 7 tests

Beats GPT-5.5, GPT-5.6 Terra here, though they rank higher overall. Trails Qwen3.8 27B, GLM-5.2, GLM-5.3-Flash here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
SWE-bench Pro (split unspecified) lab	69.2% ✎	▄█·▄·▄██···▄ beats 3/9	very hard · avg 7.6%	▇▇ 4.1%	sharp · at its level
Arena WebDev (Code Arena)	1555.5 ✎	▁▂▄▃▅█▄▇▆▄▁▁ beats 86/119	avg 981	▇ 3.1%	sharp
SWE-bench Multilingual lab	84.4% ✎	▃··▃·▅▃▃█··▃ beats 3/9	hard · avg 31%	▇ 2.0%	typical
SWE-bench Verified lab	88.6% ✎	▁▁▁▁▁▁▂▄█▆▃▁ beats 128/131	hard · avg 28%	▇ 1.9%	sharp
E2E-SWE (build a working codebase from scratch, 186 tasks) pass@1	51.9% ✎	▃··▅··▃█▃▅▃▅ beats 9/12	very hard · avg 3.1%	▇ 1.4%	sharp · at its level
Luau Leetcode fully solved (Torpedo Software)	79.8% ✎	▁▁▁▁·▁▂▃▅▃█▇ beats 33/49	hard · avg 18%	▇ 1.0%	typical
DeepSWE v1.1 (resolved) lab	59.0% ✎	█·····████·█ beats 0/5	hard · avg 18%	▇ 1.0%	at its level

chat & writing▁▁▁▃▅▄▅█▆▆▇▃ #21 of 393 * · 2.0% of its score · 2 tests

Beats GPT-6 Astra, Claude Sonnet 5.5, GPT-6 Sol here, though they rank higher overall. Trails Gemini 3.6 Flash, Claude Opus 4.6, Muse Spark 1.1 here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
LMArena (text)	1481.5 ✎	▁▁▁▂▃▃▃▆▆▇█▃ beats 359/379	avg 1268	▇ 1.8%	blunt
IFEval (aggregation unspecified)	86.0% ✎	█·██·█···█·█ beats 0/5	easy · avg 74%	▇ 0.2%	maxed out

math▁▂▃▅▆▅▄█▄▃▁▁ #17 of 398 · 37% of its score · 16 tests

Beats Qwen3.8 Max (2.4T-A95B) here, though they rank higher overall. Trails GPT-5.6 Luna, GPT-6 Luna here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
FrontierMath tier 4 v2	56.1% ✎	▁▂▁▂█▅▃▂▁▁▁▁ beats 49/63	very hard · avg 0.1%	▇▇▇ 8.2%	sharp · at its level
FrontierMath tiers 1–3 v2	80.0% ✎	▂··▁▁▃▄▄█▅▃▃ beats 68/82	very hard · avg 1.2%	▇▇▇ 6.5%	sharp
MathArena ArXivMath June 2026 (48 problems)	70.0% ✎	▂···▄·██▆▆▄▂ beats 11/19	very hard · avg 4.6%	▇▇ 4.0%	sharp · at its level
MathArena ArXivMath April 2026 (40 problems)	62.5% ✎	▁·····▅██▄·▁ beats 14/20	very hard · avg 7.2%	▇ 3.7%	sharp · at its level
MathArena ArXivMath February 2026 (32 problems)	60.2% ✎	▁▁▁▁█▇▄▃▁▃·▃ beats 21/24	very hard · avg 9.1%	▇ 2.2%	at its level
MathArena ArXivMath March 2026 (30 problems)	75.0% ✎	█▃▅·█▃▃·▅·▃▃ beats 13/14	very hard · avg 15%	▇ 1.8%	at its level
MathArena BrokenArXiv April 2026 (61 problems)	34.4% ✎	▂▂▆█▃▃▂▂▂··▂ beats 12/18	very hard · avg 0.7%	▇ 1.8%	sharp · at its level · noisy
MathArena ArXivLean June 2026 (46 problems)	23.9% ✎	▄····▄▄███·▄ beats 6/9	very hard · avg 0.7%	▇ 1.3%	sharp

8 more tests

test	result	vs the others in its run	difficulty	counts	why
MathArena ArXivMath May 2026 (40 problems)	65.0% ✎	▁··▁▄█▄▁·▃·▁ beats 12/17	very hard · avg 10%	▇ 1.3%	at its level
MathArena BrokenArXiv June 2026 (54 problems)	41.7% ✎	▄··█▂▂▃▆▄··▂ beats 8/14	very hard · avg 1.1%	▇ 1.2%	sharp · at its level · noisy
MathArena BrokenArXiv May 2026 (50 problems)	23.0% ✎	██·▄▄▄····▄▄ beats 6/8	very hard · avg 1.6%	▇ 1.2%	sharp · at its level
MathArena BrokenArXiv March 2026 (56 problems)	35.7% ✎	█·▆█▂·▃▂···▂ beats 12/14	very hard · avg 0.8%	▇ 1.1%	sharp · at its level · noisy
MathArena BrokenArXiv February 2026 (31 problems)	35.5% ✎	▃▂▂▃█▃▂·▃··▂ beats 14/16	very hard · avg 0.7%	▇ 1.1%	sharp · at its level · noisy
IMO-AnswerBench	75.3% ✎	▁▁·▁▁▁▄▅█▄▂▃ beats 14/42	hard · avg 25%	▇ 0.5%	noisy
USAMO 2026 lab	96.7% ✎	█·██······██ beats 3/4	hard · avg 16%	▇ 0.5%	sharp · noisy
OTIS mock AIME 2024–2025	98.3% ✎	▃▃▄▃▃▃▇▇█▃▃▅ beats 174/193	very hard · avg 8.5%	▇ 0.2%	sharp · maxed out · noisy

vision & documents▁▁▃▂▁▃▄▂▆▇█▄ #15 of 147 · 3.4% of its score · 4 tests

Beats Claude Sonnet 5.5, GPT-5.5, MiMo-V2.6-Pro here, though they rank higher overall. Trails Claude Opus 4.7, Muse Spark 1.2, Qwen3.8-Omni-Flash here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
Arena vision (style controlled)	1286.4 ✎	▁▁▂▂▂▂▄▅▅▆██ beats 132/145	avg 1068	▇ 1.2%	blunt
CharXiv (reasoning questions) lab	80.5% ✎	███··█·██··█ beats 4/6	hard · avg 29%	▇ 1.0%	typical
Arena document (style controlled)	1485.9 ✎	▁▁▃▃▄▃▇█▅▇▄▅ beats 31/37	avg 1335	▇ 0.6%	blunt
SciDocBench scientific-document workflows (overall, 496 instances)	53.4% ✎	▃▃···█▃▅▃·▃▅ beats 6/11	hard · avg 26%	▇ 0.6%	blunt · at its level

web search▂▁▂▆▄█▃▃▁▁·▁ #26 of 80 · 4.0% of its score · 5 tests

Trails GPT-4o mini (2024-07-18), Mistral Small 3.1 24B (2503), GPT-4o (2024-08-06) here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
Humanity's Last Exam with tools lab	57.9% ✎	▁······▁·▄·█ beats 4/10	very hard · avg 14%	▇ 1.1%	at its level
BrowseComp with context management lab	84.3% ✎	▅·▅▃·█▃·█▅·▃ beats 10/14	medium · avg 44%	▇ 0.8%	typical
Arena search systems (style controlled)	1195.0 ✎	▁·▁▁▂▂▂█▅▂▁▁ beats 19/30	avg 1055	▇ 0.7%	blunt
Legal Research Bench (agentic, all-pass)	42.9% ✎	▄█··▄▄▄██·▄█ beats 12/12	very hard · avg 5.8%	▇ 0.7%	at its level
RWSearch	77.3% ✎	████··█····█ beats 1/5	medium · avg 44%	▇ 0.6%	typical

other tests17 not in a task · 25% of its score

test	result	vs the others in its run	difficulty	counts	why
EBR-bench (Epoch)	28.6% ✎	▄▂·▆█▂▆▄▆▆·▄ beats 13/23	very hard · avg 0.3%	▇▇ 4.3%	sharp · at its level
Furniture Assembly (Epoch)	42.5% ✎	█▆▆▆▆▂▅▂▂▂·▅ beats 18/30	very hard · avg 1.2%	▇ 3.2%	sharp · at its level
Mystery Game Puzzles (Epoch)	36.0% ✎	█··▂▃▇▄█▂▃▁▁ beats 58/74	very hard · avg 9.4%	▇ 3.0%	sharp · at its level · noisy
Humanity's Last Exam lab	49.8% ✎	▂·····▂·▄▃█▄ beats 4/10	very hard · avg 5.8%	▇ 2.2%	at its level
ARC-AGI-2 public evaluation set (ARC Prize model_baseline, pass@2)	76.9% ✎	█··▂▅▁▃▁▃·▁▃ beats 23/24	very hard · avg 0.3%	▇ 2.2%	sharp · at its level · noisy
GPQA Diamond +1	91.0% ✎	▁▁▁▁▂▄▆▄▅▇█▅ beats 183/208	hard · avg 41%	▇ 2.1%	sharp · maxed out · small run · several runs
Chess Puzzles (Epoch)	34.0% ✎	█▁▁▁▁▂▁▄▃▂▁▁ beats 110/135	very hard · avg 5.9%	▇ 1.7%	sharp · at its level · noisy
Legal Research Bench (single-shot, no tools, all-pass)	8.7% ✎	▃▃█▃·▅···█▃▃ beats 12/12	very hard · avg 0.8%	▇ 1.0%	too hard for it

9 more tests

test	result	vs the others in its run	difficulty	counts	why
AgentWorldBench (open-ended, overall)	56.6% ✎	▄▄··▄▄·▄█▄██ beats 9/11	hard · avg 25%	▇ 0.9%	blunt · at its level
SimpleQA Verified	53.0% ✎	▁▁▃▂▃▃▆▆█▃▂▅ beats 59/78	very hard · avg 8.0%	▇ 0.7%	at its level
FORTE (AGI-Eval)	77.2% ✎	▄···█·····▄█ beats 3/5	medium · avg 37%	▇ 0.7%	typical
Icelandic LLM Leaderboard average (Miðeind, Sept 2026 suite)	63.9% ✎	▁▁▃▂▂▄▁▃▄▄▄█ beats 29/38	very hard · avg 14%	▇ 0.6%	at its level
WritingBench	85.2% ✎	▁·▁▂▁··▅█▂▂▁ beats 13/20	medium · avg 55%	▇ 0.6%	typical
ARC-AGI-1 public evaluation set (ARC Prize model_baseline, pass@2)	95.5% ✎	▁·▇▃▁·█▇▁·▇▃ beats 24/27	very hard · avg 4.1%	▇ 0.5%	sharp · noisy
HealthBench lab	58.8% ✎	█·▄·····▄▄▄▄ beats 0/6	hard · avg 29%	▇ 0.5%	blunt · at its level
RobloxQA v2.0 multiple choice (Torpedo Software)	96.7% ✎	▁▁▃▁█▄▃▃▆▂▂▂ beats 38/49	easy · avg 78%	▇ 0.3%	maxed out
NTCIR SciClaimEval Subtask 1 pair-accuracy (dev, single model)	67.6% ✎	▃▃···▃·▃▃█·▃ beats 4/8	medium · avg 45%	▇ 0.3%	blunt · at its level

$ where to run it

Through an API

api · $10.00/M · $5 in / $25 out per 1M tokens · 1M context · docs.anthropic.com ↗ · price sourceStandard base token rate from models.dev. Provider-specific tiers, long-context surcharges, caching and promotions may differ.

$ how the score was worked out · Epoch AI — internal evaluatio… 29%, Anthropic system cards and mo… 21%, MathArena ArXivMath / BrokenA… 21% and 14 more

Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works

source	results	share of the score	licence
Epoch AI — internal evaluations	9	▇▇▇▇▇▇▇▇▇ 29%	CC BY 4.0
Anthropic system cards and model announcements (lab-reported scores) · lab-reported	15	▇▇▇▇▇▇ 21%	Factual vendor-reported score (not licensed content)
MathArena ArXivMath / BrokenArXiv / ArXivLean model-output datasets	11	▇▇▇▇▇▇ 21%	CC BY-SA 4.0
LMArena leaderboard dataset	6	▇▇▇ 11%	CC BY 4.0
meituan-longcat/LongCat-2.0 model card	6	▇ 4%	mit
ARC Prize Foundation model_baseline outputs on ARC-AGI-1/2 public evaluation sets	2	▇ 3%	MIT
Terminal-Bench 4.0 official leaderboard submissions	1	▇ 2%	Apache-2.0
Legal Research Bench (Vals AI, arXiv:2610.00609v1)	2	▇ 2%	CC-BY-4.0
τ²/τ³-bench leaderboard: Sierra's own evaluation runs	1	▇ 1%	MIT
E2E-SWE: Benchmarking LLMs on Building Working Codebases from Scratch (arXiv:2609.38335v1)	1	▇ 1%	CC BY 4.0
Roblox LLM Leaderboard (Torpedo Software): RobloxQA v2.0 and Luau Leetcode	2	▇ 1%	MIT
Qwen/Qwen-AgentWorld-35B-A3B model card	1	▇ 1%	Apache-2.0
zai-org/GLM-5.2 model card	1	▇ 1%	mit
Miðeind Icelandic LLM Leaderboard	1	▇ 1%	CC BY 4.0
SciDocBench: A Workflow-Centered Benchmark and Data Pipeline for Scientific Document Understanding (arXiv:2609.05141v2)	1	▇ 1%	CC BY 4.0
KlinikeBench: Evaluating Language Models Beyond Diagnostic Accuracy (arXiv:2609.38480v2)	1	▇ 1%	CC BY 4.0
SciTrue: Reliable Scientific Claim Validation with Frontier and Open Language Models at the NTCIR SciClaimEval Task — authors' own single-model dev evaluation	1	▇ 0%	CC BY 4.0
prior · pull toward the middle		▇ 0%
weighted average		= 78.1