Coding · Benchmark profile
Software engineering repository tasks designed as a refreshed SWE-bench style evaluation.
Data verified 15 July 2026 · Methodology 1.2.0
Benchmark score on SWE-Rebench
Claude Opus 4.6 leads at 65.3%, followed by GLM-5 (62.8%) and GLM-5.1 (62.7%).
Visual analysis
Three views of the same sourced leaderboard: model placement, the score curve, and descriptive provider averages. These visuals never alter ranking calculations.
8 ranked models · higher is better
| Model | Rank | Provider | Score |
|---|---|---|---|
| Claude Opus 4.6 | #1 | Anthropic | 65.3% |
| GLM-5 | #2 | Z.AI | 62.8% |
| GLM-5.1 | #3 | Z.AI | 62.7% |
| Claude Sonnet 4.6 | #4 | Anthropic | 60.7% |
| Kimi K2.5 | #5 | Moonshot AI | 58.5% |
| GPT-5.3-Codex | #6 | OpenAI | 58.2% |
| MiniMax M2.7 | #7 | MiniMax | 51.9% |
| Gemma 4 31B | #8 | 41.6% |
Shows how quickly scores separate down the field
| Model | Rank | Provider | Score |
|---|---|---|---|
| Claude Opus 4.6 | #1 | Anthropic | 65.3% |
| GLM-5 | #2 | Z.AI | 62.8% |
| GLM-5.1 | #3 | Z.AI | 62.7% |
| Claude Sonnet 4.6 | #4 | Anthropic | 60.7% |
| Kimi K2.5 | #5 | Moonshot AI | 58.5% |
| GPT-5.3-Codex | #6 | OpenAI | 58.2% |
| MiniMax M2.7 | #7 | MiniMax | 51.9% |
| Gemma 4 31B | #8 | 41.6% |
Average within the visible cohort · descriptive only
| Provider | Models | Average score |
|---|---|---|
| Anthropic | 2 | 63% |
| Z.AI | 2 | 62.8% |
| Moonshot AI | 1 | 58.5% |
| OpenAI | 1 | 58.2% |
| MiniMax | 1 | 51.9% |
| 1 | 41.6% |
One best score per model · higher is better
| Rank | Model | Provider | License | Score |
|---|---|---|---|---|
| #1 | Claude Opus 4.6 claude-opus-4-6 | Anthropic | closed | 65.3% |
| #2 | GLM-5 glm-5 | Z.AI | open | 62.8% |
| #3 | GLM-5.1 glm-5-1 | Z.AI | open | 62.7% |
| #4 | Claude Sonnet 4.6 claude-sonnet-4-6 | Anthropic | closed | 60.7% |
| #5 | Kimi K2.5 kimi-k2.5 | Moonshot AI | — | 58.5% |
| #6 | GPT-5.3-Codex gpt-5.3-codex | OpenAI | closed | 58.2% |
| #7 | MiniMax M2.7 minimax-m2-7 | MiniMax | open | 51.9% |
| #8 | Gemma 4 31B gemma-4-31b | open | 41.6% |
The top of this snapshot is led by Claude Opus 4.6 at 65.3%; third place is 2.6 points behind. The top-8 spread is 23.7 points.
About SWE-Rebench
Software engineering repository tasks designed as a refreshed SWE-bench style evaluation. Results stay tied to the exact model variant and evaluation system. Multiple systems for the same model use the best published score on this page; overall Lumina scoring uses the median of ranking-eligible rows.
Open benchmark source ↗FAQ
Software engineering repository tasks designed as a refreshed SWE-bench style evaluation.
Claude Opus 4.6 by Anthropic currently leads with 65.3%.
8 models in the LuminaBench cohort have a qualifying score on this benchmark.
Yes. This family is ranking-weighted at 3.2% of its capability category. Claude Opus 4.6 is overall #25.
Related
