SWE-bench Pro benchmark
34 AI models ranked on SWE-bench Pro. Each row shows the model’s score next to its cheapest API price per 1M tokens where it has one, so you can weigh quality against cost.
Top model today
Qwen3.8 2.4T A95B (Alibaba) leads at 68%, from $2.00 per 1M input via Aihubmix.
Data via Scale AI SWE-bench Pro →Models ranked
34
on this benchmark
Top score
68%
current leader
Type
Single eval
benchmark
Updated
Oct 14, 2026
benchmark scores
Cost vs SWE-bench Pro
32 priced modelsEach labelled name links to that model’s page. Curated to the cost-efficiency frontier, top scorers and cheapest so labels stay readable; the full field is in the leaderboard table below (and via “select all”). Reasoning models (extended “thinking” before answering) carry a ring around the dot.
Models ranked on SWE-bench Pro
highest score first · 34 models| # | ||||||
|---|---|---|---|---|---|---|
| 1 | Qwen3.8 2.4T A95B |  Alibaba | 68% | $2.00 via Aihubmix | $6.00 via Aihubmix | |
| 2 | Hy4 preview | Other | 66% | $0.834 Routed quote | $2.50 Routed quote | |
| 3 | Qwen3.8-Flash-Next | Alibaba | 63% | n/a | n/a | |
| 4 | GLM 5.2 |  Zhipu | 62% | $0.05 Routed quote | $1.98 via Scx AI | |
| 5 | Qwen3.8 27B | Alibaba | 62% | $0.04 Routed quote | $1.49 via Cerebras | |
| 6 | Nex-N2.5-Pro | Other | 61% | $0.075 Routed quote | $0.25 Routed quote | |
| 7 | Laguna S 2.1 | Poolside | 59% | $0.09 Routed quote | $0.18 Routed quote | |
| 8 | MiniMax M3 |  MiniMax | 59% | $0.23 via Wandb | $0.96 via Wandb | |
| 9 | Kimi K2.6 |  Moonshot | 59% | $0.65 via Wandb | $3.40 via Novita | |
| 10 | GLM 5.1 | Zhipu | 58% | $1.05 via DeepInfra | $3.50 via DeepInfra | |
| 11 | Hy3 | Tencent | 58% | $0.132 Routed quote | $0.528 Routed quote | |
| 12 | MiMo-V2.5-Pro |  Xiaomi | 57% | $0.435 Routed quote | $0.87 Routed quote | |
| 13 | inclusionAI: Ling 3.0 Flash | Other | 57% | $0.021 Routed quote | $0.063 Routed quote | |
| 14 | Step 3.7 Flash | Other | 56% | $0.20 via DeepInfra | $1.15 via DeepInfra | |
| 15 | MiniMax M2.7 | MiniMax | 56% | $0.25 via DeepInfra | $0.55 via Pinstripes | |
| 16 | MiMo-V2.5 | Xiaomi | 56% | $0.14 Routed quote | $0.28 Routed quote | |
| 17 | Inkling Small | Thinking Machines | 56% | $0.45 via DeepInfra | $1.20 via Baseten | |
| 18 | DeepSeek V4 Pro 0423 |  DeepSeek | 55% | $0.435 via Tencent | $0.87 via Tencent | |
| 19 | MiniMax M2.5 | MiniMax | 55% | $0.27 Routed quote | $1.08 Routed quote | |
| 20 | Inkling | Thinking Machines | 54% | $0.95 via DeepInfra | $4.05 via Baseten | |
| 21 | Qwen3.6 27B | Alibaba | 54% | $0.15 via Libertai | $0.50 via Libertai | |
| 22 | Muse Glimmer 30B | Other | 51% | $0.30 via DeepInfra | $1.20 via DeepInfra | |
| 23 | Kimi K2.5 | Moonshot | 51% | $0.45 via DeepInfra | $2.25 via DeepInfra | |
| 24 | Qwen3.6 35B A3B | Alibaba | 50% | $0.05 via Sail | $0.40 via Sail | |
| 25 | Laguna M.1 | Other | 49% | $0.20 Routed quote | $0.40 Routed quote | |
| 26 | Laguna XS 2.1 | Poolside | 48% | $0.06 Routed quote | $0.12 Routed quote | |
| 27 | Qwen3 Coder Next | Alibaba | 44% | $0.12 Routed quote | $0.548 via Aihubmix | |
| 28 | North Mini Code |  Cohere | 40% | n/a | n/a | |
| 29 | Qwen3 Coder 480B A35b Instruct | Alibaba | 39% | $0.38 via Novita | $1.50 via Wandb | |
| 30 | MiniMax M2.1 | MiniMax | 37% | $0.30 via Minimax | $1.20 via Minimax | |
| 31 | Kimi K2 Instruct | Moonshot | 28% | $0.50 via DeepInfra | $2.00 via DeepInfra | |
| 32 | Qwen3 235B A22b | Alibaba | 21% | $0.18 via DeepInfra | $0.54 via DeepInfra | |
| 33 | gpt-oss-120b |  OpenAI | 16% | $0.03 via Wandb | $0.17 via DeepInfra | |
| 34 | GLM 4.6 | Zhipu | 10% | $0.50 via DeepInfra | $2.00 via DeepInfra |
Frequently asked questions
What is the SWE-bench Pro benchmark?
SWE-bench Pro (Scale AI) is a harder, contamination-resistant successor to SWE-bench, built from larger and more recent real-world software tasks across many repositories. The model must generate a patch that passes the project's tests. Scored as the percentage of tasks resolved, with frontier models still well short of saturation.
Which AI model scores highest on SWE-bench Pro?
Qwen3.8 2.4T A95B (Alibaba) leads with 68%, from $2.00 per 1M input tokens from Aihubmix.
How many models are ranked on SWE-bench Pro?
34 models carry a SWE-bench Pro score. Scores come from Hugging Face leaderboards; where a model has a listed API price it is drawn from LiteLLM and OpenRouter, and models with no current price are listed without one.
Other benchmarks
compare the same models on a different evalEvery weekday
AI moves fast. Here's your debrief.
News, analysis, tools, and more.
For people who build with AI