SWE-bench leaderboard

Real GitHub issues from open-source projects; a model must write a patch that makes the project’s tests pass. Claude Opus 5 leads with 97.00%.

Models
84
Leader
Claude Opus 5
Source
Vals AI
As of
Oct 24, 2026
  1. 1Claude Opus 597.0%
  2. 2GPT-5.6 Sol96.2%
  3. 3Grok 4.695.6%
  4. 4GPT-5.6 Terra95.4%
  5. 5GLM-5.395.4%
  6. 6Claude Fable 595.0%
  7. 7Kimi K393.4%
  8. 8GPT-5.6 Luna93.0%
  9. 9GLM-5.3-Flash92.0%
  10. 10DeepSeek-V4-Flash88.8%
SWE-bench, accuracy against cost per test
Open weightsProprietaryBest for the price
0%25%50%75%100%$.01$.1$1$10Cost per test, log scale →Claude Opus 5GLM-5.3GPT-5.6 LunaDeepSeek-V4-Flash

Full leaderboard

RankModelLabAccuracyCost / testHarnessPrice in / out
1Claude Opus 5Anthropic97.00%$1.29Mini-SWE-agent$5 / $25
3GPT-5.6 SolOpenAI96.20%$1.15Mini-SWE-agent$4 / $20
4Grok 4.6xAI95.60%$0.78Mini-SWE-agent$2 / $6
5GPT-5.6 TerraOpenAI95.40%$0.40Mini-SWE-agent$2 / $12
6GLM-5.3Z.ai95.40%$0.34Mini-SWE-agent$0.079 / $1.40
7Claude Fable 5Anthropic95.00%$2.05Mini-SWE-agent$10 / $50
8Kimi K3Moonshot AI93.40%$0.76Mini-SWE-agent$0.24 / $11
9GPT-5.6 LunaOpenAI93.00%$0.04Mini-SWE-agent$0.20 / $1.20
10GLM-5.3-FlashZ.ai92.00%$0.02Mini-SWE-agent$0.15 / $0.50
11DeepSeek-V4-FlashDeepSeek88.80%$0.01Mini-SWE-agent$0.042 / $0.18
12Claude Opus 4.8Anthropic88.60%$1.92Mini-SWE-agent$5 / $25
13Grok 4.5xAI86.60%$0.54Mini-SWE-agent$2 / $6
14Muse Spark 1.2Meta86.60%$0.55Mini-SWE-agent$1.25 / $4.25
15Qwen3.8-27BAlibaba86.00%$1.08Mini-SWE-agent$0.20 / $2.50
16Qwen3.8-MaxAlibaba85.60%$1.13Mini-SWE-agent$2 / $6
17GLM-5.2Z.ai82.80%$0.71Mini-SWE-agent$0.034 / $7
18GPT-5.5OpenAI82.60%$1.36Mini-SWE-agent$5 / $30
19Inkling-SmallThinky82.20%$0.13Mini-SWE-agent$0.45 / $1.20
20Muse Spark 1.1Meta82.00%$0.35Mini-SWE-agent$1.25 / $4.25
21Claude Opus 4.7Anthropic82.00%$2.42Mini-SWE-agent$5 / $25
22Gemini 3.7 FlashGoogle80.80%$1.44Mini-SWE-agent$0.75 / $3.75
23Gemini 3.8 FlashGoogle80.00%$2.19Mini-SWE-agent$0.82 / $4.13
24Composer 2.5—79.60%N/ACursor CLI—
25Gemini 3.6 FlashGoogle79.60%$1.19Mini-SWE-agent$0.75 / $3.75
26Claude Sonnet 5Anthropic79.60%$1.49Mini-SWE-agent$2 / $10
27Gemini 3.5 FlashGoogle78.80%$0.95Mini-SWE-agent$1.50 / $9
28Gemini 3.1 Pro PreviewGoogle78.80%$0.78Mini-SWE-agent$2 / $12
29GPT-5.4OpenAI78.20%$0.80Mini-SWE-agent$2.50 / $15
30Claude Opus 4.6Anthropic78.20%$1.22Mini-SWE-agent$5 / $25
31Kimi K2.7 CodeMoonshot AI78.20%$0.27Mini-SWE-agent$0.67 / $3.35
32GPT-5.3-CodexOpenAI78.00%$0.46Mini-SWE-agent$1.75 / $14
33InklingThinky77.60%$0.59Mini-SWE-agent$1 / $4.05
34Claude Sonnet 4.6Anthropic77.40%$1.30Mini-SWE-agent$3 / $15
35DeepSeek-V4-ProDeepSeek77.40%$0.44Mini-SWE-agent$0.66 / $1.98
36Claude Opus 4.5Anthropic76.40%$1.05Mini-SWE-agent$5 / $25
37Gemini 3 ProGoogle76.40%$0.77Mini-SWE-agent—
38GLM-5.1Z.ai76.40%$0.46Mini-SWE-agent$1.40 / $4.40
39Kimi K2.6Moonshot AI76.20%$0.27Mini-SWE-agent$0.65 / $3.41
40GPT-5.2OpenAI75.80%$1.45Mini-SWE-agent$1.75 / $14
41Gemini 3 FlashGoogle75.00%$0.28Mini-SWE-agent$0.50 / $3
42Gemini 3.5 Flash-LiteGoogle75.00%$0.31Mini-SWE-agent$0.30 / $2.50
43MiniMax-M3MiniMax75.00%$0.42Mini-SWE-agent$0.30 / $1.20
44MiniMax-M2.1 PreviewMiniMax74.80%$0.44Mini-SWE-agent$0.30 / $1.20
45Muse SparkMeta74.40%N/AMini-SWE-agent—
46MiniMax-M2.5MiniMax74.20%$0.46Mini-SWE-agent$0.30 / $1.20
47MiMo-V2.5-ProXiaomi74.00%$0.02Mini-SWE-agent$0.30 / $0.61
48MiniMax-M2.7MiniMax73.80%$0.47Mini-SWE-agent$0.30 / $1.20
49Qwen3.6-PlusAlibaba73.40%$0.87Mini-SWE-agent$0.33 / $1.95
50GPT-5.4 miniOpenAI73.00%$0.51Mini-SWE-agent$0.75 / $4.50
51Qwen3.6-Max-PreviewAlibaba72.80%$1.57Mini-SWE-agent—
52GPT-5.2-CodexOpenAI72.40%$0.31Mini-SWE-agent$1.75 / $14
53Grok 4.20xAI72.20%$0.52Mini-SWE-agent$1.25 / $2.50
54Grok 4.3xAI71.40%$0.22Mini-SWE-agent$1.25 / $2.50
55GLM-5Z.ai71.40%$0.45Mini-SWE-agent$1 / $3.20
56Qwen 3.5 PlusAlibaba71.20%$0.96Mini-SWE-agent$0.30 / $1.80
57MiMo-V2.5Xiaomi71.00%$0.01Mini-SWE-agent$0.12 / $0.24
58Kimi K2.5Moonshot AI70.00%$0.17Mini-SWE-agent$0.50 / $2.50
59Claude Sonnet 4.5Anthropic70.00%$0.81Mini-SWE-agent$3 / $15
60Qwen 3.6 27BAlibaba70.00%$0.91Mini-SWE-agent$0.25 / $2
61GPT-5.4 nanoOpenAI69.80%$0.10Mini-SWE-agent$0.20 / $1.25
62GPT-5.1OpenAI69.80%$0.71Mini-SWE-agent$1.25 / $10
63GLM-4.7Z.ai69.40%$0.20Mini-SWE-agent$0.54 / $1.98
64GPT-5OpenAI69.00%$0.27Mini-SWE-agent$1.25 / $10
65Nemotron 3 UltraNVIDIA69.00%N/AMini-SWE-agent$0.60 / $2.40
66Qwen3.7-Max-PreviewAlibaba68.80%$3.23Mini-SWE-agent$1.48 / $4.42
67DeepSeek-V3.2-ExpDeepSeek67.60%$0.67Mini-SWE-agent$0.28 / $0.42
68Claude Haiku 4.5Anthropic66.60%$0.37Mini-SWE-agent$1 / $5
69Mistral Medium 3.5Mistral AI66.40%$4.00Mini-SWE-agent$1.50 / $7.50
70Ling 3.0 FlashinclusionAI65.20%$0.03Mini-SWE-agent$0.021 / $0.063
71Qwen3.5-FlashAlibaba64.40%$0.26Mini-SWE-agent$0.065 / $0.26
72Gemini 3.1 Flash-Lite PreviewGoogle62.80%$0.11Mini-SWE-agent$0.25 / $1.50
73Devstral 2Mistral AI62.80%$0.75Mini-SWE-agent$0.40 / $2
74GPT-5 miniOpenAI60.80%$0.05Mini-SWE-agent$0.25 / $2
75Kimi K2 ThinkingMoonshot AI60.20%$0.37Mini-SWE-agent$0.60 / $2.50
76Grok 4xAI57.80%$1.42Mini-SWE-agent—
77Laguna M.1Poolside57.60%N/AMini-SWE-agent—
78Laguna XS.2Poolside55.20%N/AMini-SWE-agent—
79Gemini 2.5 ProGoogle54.40%$0.35Mini-SWE-agent$1.25 / $10
80Nemotron 3.5 LightningFireworks AI52.80%$0.03Mini-SWE-agent$0.07 / $0.20
81Grok 4 FastxAI45.40%$0.05Mini-SWE-agent—
82Grok 4.1 FastxAI41.40%$0.03Mini-SWE-agent—
83Mistral LargeMistral AI41.40%$0.32Mini-SWE-agent$0.50 / $1.50
84gpt-oss-120bOpenAI33.60%$0.17Mini-SWE-agent$0.15 / $0.60
85Command ACohere7.80%$0.21Mini-SWE-agent$0.30 / $1.50

Related leaderboards

Sources

  • Vals AI, as of Oct 24, 2026
  • Provider list prices, as of Oct 24, 2026

Every number on this page comes from these sources; models without a published number show none. Download the data. Cite as “Nexus Explorer, nexus.inc/models”.