Compare coding models with the tested setup in view.
Start with a coding result you can trace to a named test. This page brings the score, tested setup and listed model limits together so you can make a shortlist without opening every leaderboard first.
31 language models have a retained LiveBench Coding result for 2.0 · 2026-10-25 (dataset 20261025). This view shows up to 12, ordered within that metric only.
Catalog snapshot · 12 rows shownKeep the test fixed
Read the benchmark name and version before the score. The table uses one LiveBench coding metric at a time. A result from a different collection or version does not become comparable because both numbers fit on a 0–100 scale.
The source’s configuration records the tested effort or settings. A newer check date says when the result was collected; it does not establish a new evaluation. Open the result source before treating a small score difference as decisive.
Check the model you can actually call
Use the model profile to inspect its exact identity and provider routes. A family name alone does not establish that an endpoint serves the tested variant. Context and output limits below are catalog specifications, not proof of accuracy across a long repository.
For an agent, check tool calling, structured output and the harness you plan to use. A model-only coding result does not measure the complete agent. Keep a short task from your own codebase for the final check, and inspect the diff and test results it produces.
How this view is assembled
Selection: language models with a finite result for the latest version label among retained LiveBench coding metrics. One most recently checked configuration per model; at most 12 rows. Scores are ordered only within that metric and version. Missing benchmark evidence leaves the table empty rather than creating a substitute ranking.
Coding results with their tested setup
LiveBench Coding · 2.0 · 2026-10-25 (dataset 20261025)
| Model | Score | Benchmark | Tested setup | Listed model limits | Sources and dates |
|---|---|---|---|---|---|
| Anthropic: Claude Fable 5.1 | 97.7%Higher is better | LiveBench Coding2.0 · 2026-10-25 (dataset 20261025) | Effort: max · single runIndependent resultEvaluation date not reported | 1M context128K max outputTool calling listed | Benchmark result ↗Checked OpenRouter ↗Catalog updated Experiential ↗Catalog updated |
| Anthropic: Claude Opus 5.5 | 96.3%Higher is better | LiveBench Coding2.0 · 2026-10-25 (dataset 20261025) | Effort: max · single runIndependent resultEvaluation date not reported | 1M context128K max outputTool calling listed | Benchmark result ↗Checked OpenRouter ↗Catalog updated DeepInfra ↗Catalog updated |
| DeepSeek-V4.1-Flash-1017 | 96%Higher is better | LiveBench Coding2.0 · 2026-10-25 (dataset 20261025) | Effort: max · single runIndependent resultEvaluation date not reported | 1M context393.2K max outputTool calling listed | Benchmark result ↗Checked OpenRouter ↗Catalog updated Hugging Face ↗Catalog updated |
| OpenAI: GPT-6.1 Sol | 96%Higher is better | LiveBench Coding2.0 · 2026-10-25 (dataset 20261025) | Effort: max · single runIndependent resultEvaluation date not reported | 1.1M context128K max outputTool calling listed | Benchmark result ↗Checked OpenRouter ↗Catalog updated Experiential ↗Catalog updated |
| Anthropic: Claude Sonnet 5.5 | 94.7%Higher is better | LiveBench Coding2.0 · 2026-10-25 (dataset 20261025) | Effort: max · single runIndependent resultEvaluation date not reported | 1M context128K max outputTool calling listed | Benchmark result ↗Checked OpenRouter ↗Catalog updated DeepInfra ↗Catalog updated |
| OpenAI: GPT-6.1 Terra | 94.3%Higher is better | LiveBench Coding2.0 · 2026-10-25 (dataset 20261025) | Effort: max · single runIndependent resultEvaluation date not reported | 1.1M context128K max outputTool calling listed | Benchmark result ↗Checked OpenRouter ↗Catalog updated LiveBench ↗Catalog updated |
| OpenAI: GPT-6 Astra | 93.7%Higher is better | LiveBench Coding2.0 · 2026-10-25 (dataset 20261025) | Effort: max · single runIndependent resultEvaluation date not reported | 1.1M context128K max outputTool calling listed | Benchmark result ↗Checked OpenRouter ↗Catalog updated Experiential ↗Catalog updated |
| Claude Haiku 5.5 | 91.4%Higher is better | LiveBench Coding2.0 · 2026-10-25 (dataset 20261025) | Effort: max · single runIndependent resultEvaluation date not reported | — contextOutput limit not reportedTool calling not reported | Benchmark result ↗Checked LiveBench ↗Catalog updated |
| Muse Spark 1.3 | 90.2%Higher is better | LiveBench Coding2.0 · 2026-10-25 (dataset 20261025) | Effort: max · single runIndependent resultEvaluation date not reported | — contextOutput limit not reportedTool calling not reported | Benchmark result ↗Checked LiveBench ↗Catalog updated |
| Google: Gemini 4 Argon | 89.7%Higher is better | LiveBench Coding2.0 · 2026-10-25 (dataset 20261025) | Effort: high · single runIndependent resultEvaluation date not reported | 1M context943.7K max outputTool calling listed | Benchmark result ↗Checked OpenRouter ↗Catalog updated Experiential ↗Catalog updated |
| Grok 4.6 | 89.1%Higher is better | LiveBench Coding2.0 · 2026-10-25 (dataset 20261025) | Effort: xhigh · single runIndependent resultEvaluation date not reported | — contextOutput limit not reportedTool calling not reported | Benchmark result ↗Checked LiveBench ↗Catalog updated |
| Qwen3.8-27B | 86.8%Higher is better | LiveBench Coding2.0 · 2026-10-25 (dataset 20261025) | Effort: max · single runIndependent resultEvaluation date not reported | 1M context65.5K max outputTool calling listed | Benchmark result ↗Checked OpenRouter ↗Catalog updated Hugging Face ↗Catalog updated |
Where these rows come from
We join catalog specifications, recorded provider quotes and published evaluation records where exact model identities are available. Some observations come through aggregators or benchmark collections. Multiple listings do not establish independent confirmation of a claim.
Open the source beside a row to check the original evidence. A collection or check date is separate from a model release or evaluation date. Unknown values stay unknown. Recorded prices are in USD with their original billing units and settings.
See data coverage and sources →Compare for your own use case
Keep your task, model identity and provider configuration together. Explore the catalog yourself, or connect your AI to read the same source-backed records.