Skip to content
Model Menu
Models
Compare
Leaderboard
Benchmarks
⌘K
Top
Chat
Coding
Agents
Writing
Vision
Image
Video
Speech
Embeddings
Decision
Music
Filters
Best for Coding
All models
2314
Chat & Knowledge
563
Coding
383
Agents & Tool Use
543
Writing & Language
361
Vision & Documents
397
Image Generation & Editing
184
Video Generation
148
Speech & Voice
358
Embeddings & Retrieval
688
Decision models
118
Music & Sound
29
Terminal-Bench 4.0
DeepSWE 1.1
FrontierCode 1.1
SWE-bench Pro public
WebDev Arena
SWE-bench Pro v2 full
All benchmarks
51
Coding benchmarks
51
×
All
51
Coding
12
Agents
13
Code review
1
Benchmark
By
Leader
Models
Coding
12
WebDev Arena
Human pairwise preferences in the stated webdev vote pool
Arena
Claude Opus 5.5
1,807
97
+2 versions & slices
October 23, 2026
Human pairwise preferences in the stated webdev vote pool
Arena
Claude Opus 5.5
1,807
98
October 19, 2026
Human pairwise preferences in the stated webdev vote pool
Arena
Claude Opus 5.5
1,807
97
SciCode
Writing code for real scientific computing problems (physics, biology, math)
AA
Claude Opus 5.5
66.9%
142
LiveCodeBench
AA archived
AA evaluates solutions to newly collected competitive-programming problems
AA
Gemini 3 Flash Preview
90.8%
152
+2 versions & slices
official artifact (through April 2025)
Mean pass@1 over each model’s available problems in the current official public generation artifact; its latest problem is dated April 7, 2025
LiveCodeBench
o3
84.7%
10
LiveCodeBench
Reported by the lab with its own setup
Lab
DeepSeek V4 Pro 0813
93.5
8
Aider Polyglot
two attempts
Models edit code for 225 Exercism exercises in six languages, with a second attempt after test feedback
Aider
GPT-5
88%
15
+2 versions & slices
two attempts · diff edits · December 2024 run
Models edit code for 225 Exercism exercises in six languages, with a second attempt after test feedback
Aider
Qwen2.5 Coder 32B Instruct
8%
1
Aider Polyglot
Reported by the lab with its own setup
Lab
DeepSeek V3.1 Terminus
76.1
1
MirrorCode
Agents reimplement software projects by observing their behavior
Epoch
Claude Opus 5.5
77.4%
10
MultiLangBench
Models build a SQLite-compatible database engine in 20 programming languages
OpenAI
GPT-6 Astra
87.4%
8
Agents
11
Terminal-Bench
4.0
Agents use a terminal to complete executable coding, system administration and research tasks in the 4.0 suite
Claude Sonnet 5.5
63.6%
134
+2 versions & slices
2.1
Older, easier version of Terminal-Bench; Still widely reported
AA
Claude Fable 5.1
91.4%
156
3.0
Reported by the lab with its own setup
Lab
Claude Opus 5
43.3
6
DeepSWE
1.1
Agents change repository code to resolve software engineering tasks under the stated DeepSWE edition
GPT-6 Astra
74.1%
44
+1 versions & slices
1
Agents change repository code to resolve software engineering tasks under the stated DeepSWE edition
Datacurve
GPT-5.5
70%
20
FrontierCode
1.1
Agents implement changes in real repositories, graded against maintainer-defined requirements
AA
Claude Opus 5.5
57.1%
38
SWE-bench Pro (v2 full)
Agents resolve repository issues under the network-locked v2 protocol; full and 51-task Hard results are separate
Scale
Claude Opus 5
99.4%
16
SWE-bench Pro (v2 hard)
Agents resolve repository issues under the network-locked v2 protocol; full and 51-task Hard results are separate
Scale
Claude Opus 5
98%
16
SWE-bench Pro (private)
Agents resolve long-horizon professional repository tasks; public and private splits are separate
Scale
Muse Spark 1.1
51.5%
9
Terminal-Bench Science
0.1
Agents complete expert-authored research workflows across five scientific domains in a terminal
AA
GPT-6 Astra
63.3%
45
Terminal-Bench Hard
Agents solve difficult software, system-administration and data-processing tasks in a terminal
AA
GPT-5.6 Sol
65.9%
239
Code review
1
Code Review Bench
Core F2
Review systems identify pull-request defects; core F2 weights recall more heavily than precision over the published gold set
Martian
Reported by labs
25
SWE-bench Pro (edition unstated)
Reported by the lab with its own setup
Lab
Claude Opus 5.5
89.9
44
DeepSWE (edition unstated)
Reported by the lab with its own setup
Lab
Claude Fable 5
70
11
SciCode Verified pass@1 n=6
Reported by the lab with its own setup
Lab
GPT-6 Astra
94.2
11
Terminal Bench
2.0
Reported by the lab with its own setup
Lab
GPT-5.4
75.1
8
SWE-bench Verified resolved
Reported by the lab with its own setup
Lab
DeepSeek R1
49.2
4
LiveCodeBench pass@1
Reported by the lab with its own setup
Lab
DeepSeek R1 Distill Llama 70B
57.5
8
Codeforces Rating
Reported by the lab with its own setup
Lab
DeepSeek V4.1 Flash 1017
3,471
14
+1 versions & slices
max effort
Reported by the lab with its own setup
Lab
DeepSeek V4 Pro 0813
3,206
1
SWE Verified
Reported by the lab with its own setup
Lab
Claude Opus 4.6
80.8
7
ProgramBench
Reported by the lab with its own setup
Lab
Claude Opus 5
37
6
NL2Repo-Bench
Reported by the lab with its own setup
Lab
Claude Opus 5
75.3
6
Meta Internal Coding Bench
Reported by the lab with its own setup
Lab
Claude Opus 4.8
69
5
Surge Human Eval on Code
Reported by the lab with its own setup
Lab
Claude Opus 5
4.2
5
SWE Atlas QnA
Reported by the lab with its own setup
Lab
DeepSeek V4 Pro 0813
66
5
SWE Multilingual
Reported by the lab with its own setup
Lab
Claude Opus 4.6
77.5
4
Codeforces percentile
Reported by the lab with its own setup
Lab
o1 1217
96.6
4
COBOL migration eval
Reported by the lab with its own setup
Lab
GPT-6.1 Sol
86.6
4
Vibe Code Bench
v1.1
Reported by the lab with its own setup
Lab
Muse Spark 1.1
72.2
2
SWE Atlas Codebase QnA
Reported by the lab with its own setup
Lab
Muse Spark 1.1
42
2
SWE-bench Multilingual
Reported by the lab with its own setup
Lab
DeepSeek V3.1 Terminus
57.8
1
Terminal-bench version unstated
Reported by the lab with its own setup
Lab
DeepSeek V3.1 Terminus
36.7
1
Codeforces-Div1 rating
Reported by the lab with its own setup
Lab
DeepSeek V3.1 Terminus
2,046
1
Codeforces rating
effort unstated
Reported by the lab with its own setup
Lab
DeepSeek V4 Pro 0813
3,348
1
SWE Marathon
Reported by the lab with its own setup
Lab
Kimi K3.1 Pro
51.7
1
Kimi Code Bench 3.0 Long-horizon
Reported by the lab with its own setup
Lab
Kimi K3.1 Pro
61.2
1
Legacy · saturated or no longer reported
2
SWE-bench Pro (public)
Agents resolve long-horizon professional repository tasks; public and private splits are separate
Muse Spark 1.1
61.5%
19
SWE-bench Verified
Epoch
Swe Bench Verified: Agents modify real repositories to resolve the stated Verified issues
Epoch
Claude Opus 4.7
83.5%
24
No benchmark matches
Model
SWE-bench Pro v2 full
Per task
Price
Speed
Supports
Context
Claude Opus 5
Anthropic
99.4
xhigh
$4.88
$5
/
$25
52
1M
Gemini 4 Argon
Google
99.2
high
$1.62
$2
/
$10
79
1M
Claude Fable 5.1
Anthropic
99.1
high
$3.91
$10
/
$50
58
1M
Kimi K3
Moonshot AI
Open
97.7
max
$2.00
$3
/
$15
55
1M
GPT-6 Astra
OpenAI
96.9
high
$1.73
$10
/
$50
60
1.05M
Muse Wonder
Meta
96.7
xhigh
—
$4
/
$20
—
1M
GPT-6.1 Sol
OpenAI
96.1
xhigh
$0.39
$2
/
$10
63
1.05M
Claude Sonnet 5.5
Anthropic
95.8
xhigh
$2.01
$2
/
$10
110
1M
GLM 5.3
Z.ai
Open
95.6
max
$2.01
$1.40
/
$4.40
94
1M
GPT-5.6 Sol
OpenAI
95.5
xhigh
$1.18
$4
/
$20
84
1.05M
Muse Spark Open
Meta
Open
95
xhigh
$0.84
$1.25
/
$4.25
183
1M
Gemini 3.8 Flash
Google
94.9
high
$3.49
$0.75
/
$3.75
132
1M
Claude Sonnet 5
Anthropic
93.2
xhigh
$2.87
$2
/
$10
68
1M
GPT-5.6 Terra
OpenAI
92.4
xhigh
$0.63
$2
/
$12
133
1.05M
T
Inkling
Thinking Machines
Open
89.9
xhigh
$0.60
$0.95
/
$4.05
204
256K
Claude Haiku 5.5
Anthropic
87.2
high
$0.079
$0.10
/
$0.50
185
1M
Show 367 not measured on SWE-bench Pro v2 full
