SWE-bench

🏆 Leaderboard

As of August 20, 2026, DeepSeek-V4-Pro-0813 is #1 for SWE-bench at 96.4%. Ranked by the SWE-bench Verified score SWE-bench leaderboard with Verified, Pro, and DeepSWE scores plus API pricing. See which LLM actually fixes real GitHub issues.

Updated August 20, 2026282 models33 providers


DeepSeek-V4-Pro-0813OSS DeepSeek · Open Source	96.4%	—	—	62.7%	62.8%	—	$5.28
GPT-5.6 Sol OpenAI · Proprietary	96.2%	64.6%	—	72.7%	72.7%	—	$35.00
Claude Opus 5 Anthropic · Proprietary	96%	79.2%	—	73.7%	73.7%	—	$30.00
Grok 4.6 xAI · Proprietary	95.6%	—	—	—	67.5%	—	$8.00
Claude Mythos 5 Anthropic · Proprietary	95.5%	80.3%	77.8%	—	—	80.3%	$60.00
Claude Fable 5 Anthropic · Proprietary	95%	80%	—	69.9%	69.9%	80%	$60.00
Claude Mythos Preview Anthropic · Proprietary	93.9%	77.8%	87.3%	—	—	—	$60.00
Kimi K3OSS Moonshot AI · Open Source	93.4%	—	—	67.5%	68.5%	—	$18.00
GPT-5.6 Luna OpenAI · Proprietary	93%	62.7%	—	67.2%	67.2%	—	$1.40
DeepSeek-V4-Flash-0731OSS DeepSeek · Open Source	88.8%	—	—	54.4%	53.3%	—	$1.76
Claude Opus 4.8 Anthropic · Proprietary	88.6%	69.2%	84.4%	58.2%	59.0%	—	$30.00
Claude Opus 4.7 Anthropic · Proprietary	87.6%	64.3%	—	54.2%	—	—	$30.00
Muse Spark 1.2 Meta · Proprietary	86.6%	—	—	—	54.9%	—	$5.50
Grok 4.5 xAI · Proprietary	86.6%	64.7%	—	53%	53.8%	—	$8.00
Qwen3.8 MaxOSS Qwen · Open Source	85.6%	67.7%	—	—	57.5%	—	$8.00
Claude Sonnet 5 Anthropic · Proprietary	85.2%	63.2%	78.3%	53.9%	53.9%	—	$12.00
GPT-5.5 OpenAI · Proprietary	82.6%	58.6%	—	70.0%	67.0%	58.6%	$35.00
Muse Spark 1.1 Meta · Proprietary	82%	61.5%	—	53.3%	53.3%	—	$5.50
Claude Opus 4.5 Anthropic · Proprietary	80.9%	—	—	—	—	—	$30.00
Gemini 3.7 Flash Google · Proprietary	80.8%	—	—	—	65.5%	—	$4.50
Claude Opus 4.6 Anthropic · Proprietary	80.8%	—	77.8%	27.6%	—	—	$30.00
Gemini 3.1 Pro Google · Proprietary	80.6%	54.2%	—	9.7%	11.7%	—	$17.50
DeepSeek-V4-Pro-MaxOSS DeepSeek · Open Source	80.6%	55.4%	76.2%	—	—	—	$5.22
MiniMax M3OSS MiniMax · Open Source	80.5%	59%	—	20.4%	—	—	$1.50
Qwen3.7 Max Qwen · Proprietary	80.4%	60.6%	78.3%	17.7%	—	—	$5.00
MiniMax M2.5OSS MiniMax · Open Source	80.2%	55.4%	—	—	—	—	$1.50
Kimi K2.6OSS Moonshot AI · Open Source	80.2%	58.6%	76.7%	23.9%	—	—	$4.93
Inkling-SmallOSS Thinking Machines · Open Source · via Thinking Machines Lab	80.2%	55.9%	—	—	—	—	$1.50
GPT-5.2 OpenAI · Proprietary	80%	79.9%	—	—	—	—	$15.75
Gemini 3.6 Flash Google · Proprietary	79.6%	58.7%	—	48.6%	46.7%	58.7%	$4.50
Claude Sonnet 4.6 Anthropic · Proprietary	79.6%	—	—	31.8%	29.9%	—	$18.00
Gemini 3.5 Flash Google · Proprietary	79.3%	55.1%	—	28.3%	36.1%	—	$10.50
DeepSeek-V4-Flash-MaxOSS DeepSeek · Open Source	79%	52.6%	73.3%	—	—	—	$0.42
MiMo-V2.5-ProOSS Xiaomi · Open Source	78.9%	57.2%	—	19.5%	—	—	$1.30
Qwen3.6 Plus Qwen · Proprietary	78.8%	56.6%	73.8%	2.6%	—	—	$3.50
GLM-5.2OSS Z AI · Open Source	78.7%	62.1%	—	41.5%	43.8%	—	$5.80
DeepSeek-V4-Flash-0423OSS DeepSeek · Open Source	78.6%	52.3%	70.2%	—	—	—	$0.30
Kimi K2.7 CodeOSS Moonshot AI · Open Source	78.2%	—	—	30.5%	30.5%	—	$4.93
Hy3OSS Tencent · Open Source	78%	57.9%	75.8%	28%	—	—	$0.66
Gemini 3 Flash Google · Proprietary	78%	—	—	5.1%	—	—	$3.50
GLM-5OSS Z AI · Open Source	77.8%	—	—	—	—	—	$4.20
Qwen3.7-Plus Qwen · Proprietary	77.7%	57.6%	75.8%	—	—	—	$1.60
Mistral Medium 3.5OSS Mistral · Open Source · via Mistral AI	77.6%	—	—	—	—	—	$9.00
Qwen3.6-27BOSS Qwen · Open Source	77.2%	53.5%	71.3%	—	—	—	$4.20
GPT-5.4 OpenAI · Proprietary	76.9%	57.7%	—	55.5%	51.8%	88.4%	$17.50
Kimi K2.5OSS Moonshot AI · Open Source	76.8%	50.7%	73%	—	—	—	$3.68
Seed 2.0 Pro ByteDance · Proprietary	76.5%	—	—	—	—	—	$3.50
Qwen3.5-397B-A17BOSS Qwen · Open Source	76.4%	—	69.3%	—	—	—	$4.20
GPT-5.1 Thinking OpenAI · Proprietary	76.3%	—	—	—	—	—	$11.25
GPT-5.1 Instant OpenAI · Proprietary	76.3%	—	—	—	—	—	$11.25

Showing 1–50 of 282 models

Next step

You found the model. Now ship the product.

Auth, billing, and the API layer are already decided. Fork 8 finished AI apps, or have us build the first version with you.

Get the stack — $249 Have us build it with you

The index

All Large Language Models

282 models across 33 providers. Search or jump to a lab — every model page stays linked here.

Cursor

2 models

Composer 2 Composer 2 Fast

xAI

15 models

Grok 4.6 Grok 4.5 Grok 4.3 Grok-4.1 Grok-4.1 Fast Non-Reasoning Grok-4.1 Fast Reasoning Grok-4.1 Thinking Grok 4 Fast Grok Code Fast 1 Grok-4 Fast Non-Reasoning Grok-4 Fast Reasoning Grok-4 Grok-3 Grok-3 Mini Grok-2

DeepSeek

16 models

DeepSeek-V4-Pro-0813 DeepSeek-V4-Flash-0731 DeepSeek-V4-Flash-0423 DeepSeek-V4-Flash-Max DeepSeek-V4-Pro-Max DeepSeek-V3.2 DeepSeek-V3.2 (Non-thinking)DeepSeek-V3.2-Exp DeepSeek-R1-0528 DeepSeek-V3 0324 DeepSeek R1 Distill Llama 70B DeepSeek R1 Distill Qwen 32B DeepSeek-R1 DeepSeek-V3.1 DeepSeek-V3 DeepSeek-V2.5

MiniMax

6 models

MiniMax M3 MiniMax M2.7 MiniMax M2.5 MiniMax M2.1 MiniMax M2 MiniMax M1 80K

AI21 Labs

2 models

Jamba 1.5 Large Jamba 1.5 Mini

Amazon

5 models

Nova 2 Lite Nova 2 Sonic Nova Lite Nova Micro Nova Pro

Baidu

1 models

ERNIE 4.5

ByteDance

2 models

Seed 2.1 Turbo Seed 2.0 Pro

Cohere

3 models

North Mini Code 1.0 Command A+Command R+

IBM

1 models

Granite 3.3 8B Instruct

Inception

1 models

Mercury 2

inclusionAI

1 models

Ling-3.0-flash

LG AI Research

1 models

K-EXAONE-236B-A23B

Meituan

4 models

LongCat-Flash-Lite LongCat-Flash-Thinking-2601 LongCat-Flash-Thinking LongCat-Flash-Chat

Microsoft

5 models

MAI-Code-1.1-Flash Phi 4 Mini Phi-4-multimodal-instruct Phi 4 Phi-3.5-mini-instruct

Moonshot AI

9 models

Kimi K3 Kimi K2.7 Code Kimi K2.7 Code HighSpeed Kimi K2.6 Kimi K2.5 Kimi K2 0905 Kimi K2-Instruct-0905 Kimi K2-Thinking-0905 Kimi K2 Instruct

Nous Research

1 models

Hermes 3 70B

NVIDIA

5 models

Nemotron 3.5 Lightning (30B A3B)Nemotron 3 Ultra (550B A55B)Nemotron 3 Super (120B A12B)Nemotron 3 Nano (30B A3B)Nemotron Nano 9B v2

Poolside

2 models

Laguna S 2.1 Laguna XS 2.1

Sakana AI

1 models

Sakana Namazu

StepFun

1 models

Step-3.5-Flash

Tencent

1 models

Hy3

Thinking Machines

1 models

Inkling-Small

Unisound

1 models

Upstage

1 models

Solar Pro 4

Xiaomi

3 models

MiMo-V2.5-Pro MiMo-V2.5 MiMo-V2-Flash

Which model leads SWE-bench right now?

As of August 20, 2026, DeepSeek-V4-Pro-0813 by DeepSeek is #1 for SWE-bench at 96.4%. Ranked by the SWE-bench Verified score This board also tracks SWE-bench Verified, SWE-bench Pro, SWE-bench Multilingual, DeepSWE. Next on the same board: GPT-5.6 Sol and Claude Opus 5. Related leaders: Gemini 3 Pro on SWE-bench Pro at 84.8%; Claude Mythos Preview on SWE-bench Multilingual at 87.3%. This swe-bench leaderboard ranks models by SWE-bench Verified. Scores come from public evals. Prices are the live API rates in the table above.

Sources: SWE-bench Verified (https://www.swebench.com/verified.html); OpenAI API pricing (https://developers.openai.com/api/docs/pricing); Anthropic Claude API pricing (https://platform.claude.com/docs/en/about-claude/pricing); Gemini API pricing (https://ai.google.dev/gemini-api/docs/pricing)

Top 8 for swe-bench. Ranked by the SWE-bench Verified score Input and output are dollars per million tokens.
Rank	Model	SWE-bench Verified	Input /M	Output /M
1	DeepSeek-V4-Pro-0813	96.4%	$1.32	$3.96
2	GPT-5.6 Sol	96.2%	$5.00	$30.00
3	Claude Opus 5	96%	$5.00	$25.00
4	Grok 4.6	95.6%	$2.00	$6.00
5	Claude Mythos 5	95.5%	$10.00	$50.00
6	Claude Fable 5	95%	$10.00	$50.00
7	Claude Mythos Preview	93.9%	$10.00	$50.00
8	Kimi K3	93.4%	$3.00	$15.00

Related LLM leaderboards

LLM Leaderboard Coding Math Agentic Reasoning Writing Science Finance Long Context Vision Open Source

Individual benchmark pages

Rank one eval at a time. All LLM benchmarks.

HLE Humanity's Last Exam (No Tools)Humanity's Last Exam (With Tools)Terminal-Bench 2.0 Terminal-Bench Terminal-Bench 2.1 Terminal-Bench 2 Terminal-Bench 3.0 ARC-AGI-1 Verified Chatbot Arena

SWE-bench FAQ

Who ranks #1 on the SWE-bench leaderboard?

As of August 20, 2026, DeepSeek-V4-Pro-0813 by DeepSeek ranks #1 on SWE-bench Verified at 96.4%. API pricing is $1.32/M input and $3.96/M output.

What are the top models on SWE-bench Verified?

The current SWE-bench Verified ranking as of August 20, 2026 is 1. DeepSeek-V4-Pro-0813 at 96.4%; 2. GPT-5.6 Sol at 96.2%; 3. Claude Opus 5 at 96%.

Which swe-bench model is the cheapest?

Qwen2.5-Coder 32B Instruct is the cheapest scored model on this swe-bench leaderboard at $0.09/M input and $0.09/M output ($0.18 blended). DeepSeek-V4-Pro-0813 still leads SWE-bench Verified at 96.4%.

Should I always pick the #1 SWE-bench Verified model?

Not automatically. DeepSeek-V4-Pro-0813 leads SWE-bench Verified, but a cheaper scored model can be the better production choice if the quality gap is small. Use the table to weigh SWE-bench Verified against input/output price, context window, and related evals.

How often is the SWE-bench leaderboard updated?

Scores and API prices on this page are refreshed from published evals and provider rates. The snapshot is labeled August 20, 2026. Treat it as a current index, not a one-off blog post.

What is SWE-bench?

SWE-bench asks a model to resolve real GitHub issues by editing a repository until the tests pass. SWE-bench Verified is the human-checked subset.

What is the difference between SWE-bench Verified and SWE-bench Pro?

Verified is the standard public subset. Pro is a harder set of longer, messier engineering tasks. DeepSWE is a separate agent-style software eval.

Where can I see SWE-bench with prices?

Here. Official SWE-bench sites rank accuracy only. This table adds input and output token prices so you can compare a 70% model at $1/M versus a 75% model at $15/M.

You found the model. Now ship the product.

All Large Language Models

OpenAI

Anthropic

Google

Cursor

xAI

DeepSeek

MiniMax

Qwen

AI21 Labs

Amazon

Baidu

ByteDance

Cohere

IBM

Inception

inclusionAI

LG AI Research

Meituan

Meta

Microsoft

Mistral

Moonshot AI

Nous Research

NVIDIA

Poolside

Sakana AI

StepFun

Tencent

Thinking Machines

Unisound

Upstage

Xiaomi

Z AI

Which model leads SWE-bench right now?

Related LLM leaderboards

Individual benchmark pages

SWE-bench FAQ

Who ranks #1 on the SWE-bench leaderboard?

What are the top models on SWE-bench Verified?

Which swe-bench model is the cheapest?

Should I always pick the #1 SWE-bench Verified model?

How often is the SWE-bench leaderboard updated?

What is SWE-bench?

What is the difference between SWE-bench Verified and SWE-bench Pro?

Where can I see SWE-bench with prices?