v2: add third-party benchmark figures (18 Nov 2026)
Browse files- CHANGELOG.md +6 -0
- CITATION.bib +2 -2
- DATA_DICTIONARY.md +16 -0
- README.md +12 -4
- SCORING.md +2 -0
- third_party_benchmarks.csv +79 -0
CHANGELOG.md
CHANGED
|
@@ -1,5 +1,11 @@
|
|
| 1 |
# Changelog
|
| 2 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
## v1
|
| 4 |
|
| 5 |
Initial publication prepared from TGK research records.
|
|
|
|
| 1 |
# Changelog
|
| 2 |
|
| 3 |
+
## v2 (18 November 2026)
|
| 4 |
+
|
| 5 |
+
- Added `third_party_benchmarks.csv` with the benchmark figures cited in the article update of 18 November 2026: Artificial Analysis Coding Agent Index v2.1, Runta FrontierHarness v1.1 (overall and per-model) and Joel Niklaus's Harness Leaderboard v0.1.8.
|
| 6 |
+
- Tiers, ranks and licence records are unchanged.
|
| 7 |
+
- The FrontierHarness v1.0 figures (Kimi K3, September 2026) cited in v1 of the article were replaced by v1.1 in the article; they were never part of this dataset.
|
| 8 |
+
|
| 9 |
## v1
|
| 10 |
|
| 11 |
Initial publication prepared from TGK research records.
|
CITATION.bib
CHANGED
|
@@ -2,8 +2,8 @@
|
|
| 2 |
author = {These Guys Know},
|
| 3 |
title = {TGK AI Coding Harnesses},
|
| 4 |
year = {2026},
|
|
|
|
| 5 |
url = {https://theseguysknow.io/best-ai-coding-harnesses/},
|
| 6 |
-
doi = {10.5281/zenodo.
|
| 7 |
note = {Version archived on Zenodo}
|
| 8 |
}
|
| 9 |
-
|
|
|
|
| 2 |
author = {These Guys Know},
|
| 3 |
title = {TGK AI Coding Harnesses},
|
| 4 |
year = {2026},
|
| 5 |
+
version = {v2},
|
| 6 |
url = {https://theseguysknow.io/best-ai-coding-harnesses/},
|
| 7 |
+
doi = {10.5281/zenodo.24348696},
|
| 8 |
note = {Version archived on Zenodo}
|
| 9 |
}
|
|
|
DATA_DICTIONARY.md
CHANGED
|
@@ -36,3 +36,19 @@ Each CSV row has its own source locator where the evidence supplied one. `not re
|
|
| 36 |
- `access_route`: Service or interface used. Unit: none; provider/interface name or not recorded.
|
| 37 |
- `settings`: Recorded model or run settings. Unit: none; source text or not recorded.
|
| 38 |
- `source_url`: Public source article or primary licence link. Unit: none; URL or not recorded.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 36 |
- `access_route`: Service or interface used. Unit: none; provider/interface name or not recorded.
|
| 37 |
- `settings`: Recorded model or run settings. Unit: none; source text or not recorded.
|
| 38 |
- `source_url`: Public source article or primary licence link. Unit: none; URL or not recorded.
|
| 39 |
+
|
| 40 |
+
## third_party_benchmarks.csv
|
| 41 |
+
|
| 42 |
+
- `source`: Organisation or person that published the figure. Unit: none.
|
| 43 |
+
- `benchmark`: Benchmark or leaderboard name. Unit: none.
|
| 44 |
+
- `benchmark_version`: Version of the benchmark or leaderboard. Unit: none.
|
| 45 |
+
- `checked_at`: Date TGK read the figure. Unit: calendar date; ISO date.
|
| 46 |
+
- `harness`: Harness or configuration name as published. Unit: none.
|
| 47 |
+
- `model`: Model or models behind the figure. Unit: none.
|
| 48 |
+
- `metric`: What `value` measures. Unit: none; index score (0-100), pass rate or overall score.
|
| 49 |
+
- `value`: Published figure. Unit: per `metric`; pass rates in percent.
|
| 50 |
+
- `cost_usd`: Published cost. Unit: US dollars, or not recorded.
|
| 51 |
+
- `cost_unit`: What the cost covers. Unit: none.
|
| 52 |
+
- `derived_by_tgk`: Whether TGK computed the figure from published records. Unit: none; no, or yes with method.
|
| 53 |
+
- `source_url`: Public source. Unit: none; URL.
|
| 54 |
+
|
README.md
CHANGED
|
@@ -5,13 +5,15 @@ pretty_name: TGK AI Coding Harnesses
|
|
| 5 |
|
| 6 |
Article: https://theseguysknow.io/best-ai-coding-harnesses/
|
| 7 |
|
| 8 |
-
Version:
|
| 9 |
|
| 10 |
-
18 tier-list entries, including the numbered Top 10. Licence facts for the ranked ten are taken from the TGK snapshot
|
|
|
|
|
|
|
| 11 |
|
| 12 |
## Limitations
|
| 13 |
|
| 14 |
-
This is an editorial preference dataset, not a controlled coding benchmark.
|
| 15 |
|
| 16 |
## Results
|
| 17 |
|
|
@@ -36,8 +38,14 @@ This is an editorial preference dataset, not a controlled coding benchmark. Lice
|
|
| 36 |
| 2026-10-05 | | Qwen Code | E |
|
| 37 |
| 2026-10-05 | | Antigravity | E |
|
| 38 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 39 |
## Dataset mirrors and citation
|
| 40 |
|
| 41 |
Hugging Face: https://huggingface.co/datasets/These-Guys-Know/tgk-ai-coding-harnesses
|
| 42 |
Kaggle: https://www.kaggle.com/datasets/mikehazard/tgk-ai-coding-harnesses
|
| 43 |
-
Zenodo: https://doi.org/10.5281/zenodo.
|
|
|
|
| 5 |
|
| 6 |
Article: https://theseguysknow.io/best-ai-coding-harnesses/
|
| 7 |
|
| 8 |
+
Version: v2 (18 November 2026). Zenodo DOI: https://doi.org/10.5281/zenodo.24348696. Version 1: https://doi.org/10.5281/zenodo.24316605.
|
| 9 |
|
| 10 |
+
18 tier-list entries, including the numbered Top 10, unchanged since 5 October 2026. Licence facts for the ranked ten are taken from the TGK snapshot; later checks for the other eight are dated 17 November 2026.
|
| 11 |
+
|
| 12 |
+
New in v2: `third_party_benchmarks.csv`, the 78 benchmark figures the article cites from Artificial Analysis Coding Agent Index v2.1, Runta FrontierHarness v1.1 and Joel Niklaus's Harness Leaderboard v0.1.8, all checked on 18 November 2026. The FrontierHarness per-model pass rates were computed by TGK from Runta's published trial records and are marked as derived.
|
| 13 |
|
| 14 |
## Limitations
|
| 15 |
|
| 16 |
+
This is an editorial preference dataset, not a controlled coding benchmark. The tiers come from TGK group experience. Benchmark figures and licence data are third-party information linked to their sources and are not TGK research. Benchmark scores from different sources use different tasks, models and settings, so they should not be averaged or compared across sources.
|
| 17 |
|
| 18 |
## Results
|
| 19 |
|
|
|
|
| 38 |
| 2026-10-05 | | Qwen Code | E |
|
| 39 |
| 2026-10-05 | | Antigravity | E |
|
| 40 |
|
| 41 |
+
## Files
|
| 42 |
+
|
| 43 |
+
- `model_results.csv`, `evidence_records.csv`: tiers, ranks and licence checks (unchanged from v1).
|
| 44 |
+
- `third_party_benchmarks.csv`: cited benchmark figures (new in v2).
|
| 45 |
+
- `test_prompts.csv`: empty; no prompts were used for the editorial ranking.
|
| 46 |
+
|
| 47 |
## Dataset mirrors and citation
|
| 48 |
|
| 49 |
Hugging Face: https://huggingface.co/datasets/These-Guys-Know/tgk-ai-coding-harnesses
|
| 50 |
Kaggle: https://www.kaggle.com/datasets/mikehazard/tgk-ai-coding-harnesses
|
| 51 |
+
Zenodo: https://doi.org/10.5281/zenodo.24348696
|
SCORING.md
CHANGED
|
@@ -2,4 +2,6 @@
|
|
| 2 |
|
| 3 |
Editorial ranking and tier list based on TGK group experience; no numerical benchmark scores or weights are inferred. The numbered Top 10 and the 18-tool tier list are distinct.
|
| 4 |
|
|
|
|
|
|
|
| 5 |
Source article and method: https://theseguysknow.io/best-ai-coding-harnesses/
|
|
|
|
| 2 |
|
| 3 |
Editorial ranking and tier list based on TGK group experience; no numerical benchmark scores or weights are inferred. The numbered Top 10 and the 18-tool tier list are distinct.
|
| 4 |
|
| 5 |
+
Benchmark figures in `third_party_benchmarks.csv` are reproduced from their sources as checked on 18 November 2026 and do not feed the tiers. FrontierHarness per-model pass rates were computed by TGK as verifier passes divided by scored attempts, with attempts flagged for reward hacking counted as fails, from Runta's published trial records; the overall pass rates this produces match Runta's final table.
|
| 6 |
+
|
| 7 |
Source article and method: https://theseguysknow.io/best-ai-coding-harnesses/
|
third_party_benchmarks.csv
ADDED
|
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
source,benchmark,benchmark_version,checked_at,harness,model,metric,value,cost_usd,cost_unit,derived_by_tgk,source_url
|
| 2 |
+
Artificial Analysis,Coding Agent Index,v2.1,2026-11-18,Claude Code,Claude Fable 5.2 (xhigh),index score (0-100),58.5,15.47,per task (API token prices),no,https://artificialanalysis.ai/agents/coding-agents
|
| 3 |
+
Artificial Analysis,Coding Agent Index,v2.1,2026-11-18,Devin Fusion CLI,Claude Opus 5.5 and GPT-6 Luna,index score (0-100),58.4,8.75,per task (API token prices),no,https://artificialanalysis.ai/agents/coding-agents
|
| 4 |
+
Artificial Analysis,Coding Agent Index,v2.1,2026-11-18,Claude Code,Claude Fable 5.2 (high),index score (0-100),58.2,7.51,per task (API token prices),no,https://artificialanalysis.ai/agents/coding-agents
|
| 5 |
+
Artificial Analysis,Coding Agent Index,v2.1,2026-11-18,Claude Code,Claude Opus 5.5 (max),index score (0-100),58.2,12.61,per task (API token prices),no,https://artificialanalysis.ai/agents/coding-agents
|
| 6 |
+
Artificial Analysis,Coding Agent Index,v2.1,2026-11-18,Claude Code,Claude Sonnet 5.5 (max),index score (0-100),56.3,11.40,per task (API token prices),no,https://artificialanalysis.ai/agents/coding-agents
|
| 7 |
+
Artificial Analysis,Coding Agent Index,v2.1,2026-11-18,Antigravity CLI,Gemini 4 Carbon (high),index score (0-100),55.5,9.75,per task (API token prices),no,https://artificialanalysis.ai/agents/coding-agents
|
| 8 |
+
Artificial Analysis,Coding Agent Index,v2.1,2026-11-18,Codex,GPT-6.2 Astra (max),index score (0-100),53.5,5.12,per task (API token prices),no,https://artificialanalysis.ai/agents/coding-agents
|
| 9 |
+
Artificial Analysis,Coding Agent Index,v2.1,2026-11-18,Codex,GPT-6.2 Astra (xhigh),index score (0-100),53.1,3.20,per task (API token prices),no,https://artificialanalysis.ai/agents/coding-agents
|
| 10 |
+
Artificial Analysis,Coding Agent Index,v2.1,2026-11-18,Codex,GPT-6.1 Sol (max),index score (0-100),52.2,1.51,per task (API token prices),no,https://artificialanalysis.ai/agents/coding-agents
|
| 11 |
+
Artificial Analysis,Coding Agent Index,v2.1,2026-11-18,Codex,GPT-6.2 Sol (xhigh),index score (0-100),49.8,0.90,per task (API token prices),no,https://artificialanalysis.ai/agents/coding-agents
|
| 12 |
+
Artificial Analysis,Coding Agent Index,v2.1,2026-11-18,Kimi Code CLI,Kimi K3.1 Pro,index score (0-100),48.6,2.94,per task (API token prices),no,https://artificialanalysis.ai/agents/coding-agents
|
| 13 |
+
Artificial Analysis,Coding Agent Index,v2.1,2026-11-18,Grok Build,Grok 4.7 (xhigh),index score (0-100),41.6,15.35,per task (API token prices),no,https://artificialanalysis.ai/agents/coding-agents
|
| 14 |
+
Artificial Analysis,Coding Agent Index,v2.1,2026-11-18,DeepSeek Harness,DeepSeek V4.1 Flash (max),index score (0-100),36.8,0.32,per task (API token prices),no,https://artificialanalysis.ai/agents/coding-agents
|
| 15 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,ZCode,"all five models (GLM-5.3, GLM-5.3-Flash, DeepSeek-V4.1-Flash-1017, Muse Spark Open, Kimi-K3)",pass rate,61.9,1.06,per pass (list price x tokens),no,https://frontierharness.org/leaderboard/v1.1/
|
| 16 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Codex CLI,"all five models (GLM-5.3, GLM-5.3-Flash, DeepSeek-V4.1-Flash-1017, Muse Spark Open, Kimi-K3)",pass rate,61.1,0.94,per pass (list price x tokens),no,https://frontierharness.org/leaderboard/v1.1/
|
| 17 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,KimiCode,"all five models (GLM-5.3, GLM-5.3-Flash, DeepSeek-V4.1-Flash-1017, Muse Spark Open, Kimi-K3)",pass rate,60.2,1.09,per pass (list price x tokens),no,https://frontierharness.org/leaderboard/v1.1/
|
| 18 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Grok Build,"all five models (GLM-5.3, GLM-5.3-Flash, DeepSeek-V4.1-Flash-1017, Muse Spark Open, Kimi-K3)",pass rate,60.0,1.20,per pass (list price x tokens),no,https://frontierharness.org/leaderboard/v1.1/
|
| 19 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Hermes Agent,"all five models (GLM-5.3, GLM-5.3-Flash, DeepSeek-V4.1-Flash-1017, Muse Spark Open, Kimi-K3)",pass rate,59.9,1.16,per pass (list price x tokens),no,https://frontierharness.org/leaderboard/v1.1/
|
| 20 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Claude Code,"all five models (GLM-5.3, GLM-5.3-Flash, DeepSeek-V4.1-Flash-1017, Muse Spark Open, Kimi-K3)",pass rate,58.9,1.55,per pass (list price x tokens),no,https://frontierharness.org/leaderboard/v1.1/
|
| 21 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Pi,"all five models (GLM-5.3, GLM-5.3-Flash, DeepSeek-V4.1-Flash-1017, Muse Spark Open, Kimi-K3)",pass rate,57.7,0.98,per pass (list price x tokens),no,https://frontierharness.org/leaderboard/v1.1/
|
| 22 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Maka,"all five models (GLM-5.3, GLM-5.3-Flash, DeepSeek-V4.1-Flash-1017, Muse Spark Open, Kimi-K3)",pass rate,57.7,1.13,per pass (list price x tokens),no,https://frontierharness.org/leaderboard/v1.1/
|
| 23 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,DeepSeek Harness Minimal,"all five models (GLM-5.3, GLM-5.3-Flash, DeepSeek-V4.1-Flash-1017, Muse Spark Open, Kimi-K3)",pass rate,57.4,1.15,per pass (list price x tokens),no,https://frontierharness.org/leaderboard/v1.1/
|
| 24 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Oh My Pi,"all five models (GLM-5.3, GLM-5.3-Flash, DeepSeek-V4.1-Flash-1017, Muse Spark Open, Kimi-K3)",pass rate,57.3,0.98,per pass (list price x tokens),no,https://frontierharness.org/leaderboard/v1.1/
|
| 25 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,OpenCode v2,"all five models (GLM-5.3, GLM-5.3-Flash, DeepSeek-V4.1-Flash-1017, Muse Spark Open, Kimi-K3)",pass rate,57.3,0.98,per pass (list price x tokens),no,https://frontierharness.org/leaderboard/v1.1/
|
| 26 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,DeepSeek Harness Creator,"all five models (GLM-5.3, GLM-5.3-Flash, DeepSeek-V4.1-Flash-1017, Muse Spark Open, Kimi-K3)",pass rate,56.1,1.27,per pass (list price x tokens),no,https://frontierharness.org/leaderboard/v1.1/
|
| 27 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,DeepSeek Harness Standard,"all five models (GLM-5.3, GLM-5.3-Flash, DeepSeek-V4.1-Flash-1017, Muse Spark Open, Kimi-K3)",pass rate,55.0,1.27,per pass (list price x tokens),no,https://frontierharness.org/leaderboard/v1.1/
|
| 28 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Antigravity CLI,"all five models (GLM-5.3, GLM-5.3-Flash, DeepSeek-V4.1-Flash-1017, Muse Spark Open, Kimi-K3)",pass rate,55.0,1.63,per pass (list price x tokens),no,https://frontierharness.org/leaderboard/v1.1/
|
| 29 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Codex CLI,GLM-5.3,pass rate,57.7,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 30 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Codex CLI,GLM-5.3-Flash,pass rate,59.3,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 31 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Codex CLI,DeepSeek-V4.1-Flash-1017,pass rate,67.3,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 32 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Codex CLI,Muse Spark Open,pass rate,57.7,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 33 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Codex CLI,Kimi-K3,pass rate,63.7,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 34 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,ZCode,GLM-5.3,pass rate,58.9,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 35 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,ZCode,GLM-5.3-Flash,pass rate,60.1,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 36 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,ZCode,DeepSeek-V4.1-Flash-1017,pass rate,69.0,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 37 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,ZCode,Muse Spark Open,pass rate,57.7,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 38 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,ZCode,Kimi-K3,pass rate,64.1,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 39 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Hermes Agent,GLM-5.3,pass rate,57.7,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 40 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Hermes Agent,GLM-5.3-Flash,pass rate,60.9,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 41 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Hermes Agent,DeepSeek-V4.1-Flash-1017,pass rate,62.5,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 42 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Hermes Agent,Muse Spark Open,pass rate,57.7,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 43 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Hermes Agent,Kimi-K3,pass rate,60.9,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 44 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Grok Build,GLM-5.3,pass rate,56.5,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 45 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Grok Build,GLM-5.3-Flash,pass rate,55.6,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 46 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Grok Build,DeepSeek-V4.1-Flash-1017,pass rate,66.1,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 47 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Grok Build,Muse Spark Open,pass rate,59.7,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 48 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Grok Build,Kimi-K3,pass rate,62.1,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 49 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Claude Code,GLM-5.3,pass rate,54.4,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 50 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Claude Code,GLM-5.3-Flash,pass rate,56.9,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 51 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Claude Code,DeepSeek-V4.1-Flash-1017,pass rate,68.1,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 52 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Claude Code,Muse Spark Open,pass rate,52.8,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 53 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Claude Code,Kimi-K3,pass rate,62.1,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 54 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Pi,GLM-5.3,pass rate,56.5,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 55 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Pi,GLM-5.3-Flash,pass rate,58.1,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 56 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Pi,DeepSeek-V4.1-Flash-1017,pass rate,61.3,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 57 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Pi,Muse Spark Open,pass rate,52.0,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 58 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Pi,Kimi-K3,pass rate,60.9,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 59 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Oh My Pi,GLM-5.3,pass rate,54.4,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 60 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Oh My Pi,GLM-5.3-Flash,pass rate,58.1,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 61 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Oh My Pi,DeepSeek-V4.1-Flash-1017,pass rate,57.7,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 62 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Oh My Pi,Muse Spark Open,pass rate,58.9,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 63 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,Oh My Pi,Kimi-K3,pass rate,57.7,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 64 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,DeepSeek Harness Minimal,GLM-5.3,pass rate,55.2,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 65 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,DeepSeek Harness Minimal,GLM-5.3-Flash,pass rate,55.2,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 66 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,DeepSeek Harness Minimal,DeepSeek-V4.1-Flash-1017,pass rate,63.7,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 67 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,DeepSeek Harness Minimal,Muse Spark Open,pass rate,53.6,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 68 |
+
Runta FrontierHarness,FrontierHarness Eval,v1.1,2026-11-18,DeepSeek Harness Minimal,Kimi-K3,pass rate,59.3,not recorded,not recorded,yes: computed from Runta published trial records; totals match the official table,https://frontierharness.org/leaderboard/v1.1/
|
| 69 |
+
Joel Niklaus Harness Leaderboard,Harness Leaderboard,v0.1.8,2026-11-18,OpenCode,"open models (GLM-5.3-Flash, DeepSeek-V4.1-Flash, Flint, Ripple-65B-Preview)",overall score,52.5,1.11,per solved task,no,https://joelniklaus-harness-leaderboard.hf.space/
|
| 70 |
+
Joel Niklaus Harness Leaderboard,Harness Leaderboard,v0.1.8,2026-11-18,Kimi Code,"open models (GLM-5.3-Flash, DeepSeek-V4.1-Flash, Flint, Ripple-65B-Preview)",overall score,51.7,1.37,per solved task,no,https://joelniklaus-harness-leaderboard.hf.space/
|
| 71 |
+
Joel Niklaus Harness Leaderboard,Harness Leaderboard,v0.1.8,2026-11-18,Grok Build,"open models (GLM-5.3-Flash, DeepSeek-V4.1-Flash, Flint, Ripple-65B-Preview)",overall score,50.3,1.57,per solved task,no,https://joelniklaus-harness-leaderboard.hf.space/
|
| 72 |
+
Joel Niklaus Harness Leaderboard,Harness Leaderboard,v0.1.8,2026-11-18,DeepSeek Harness,"open models (GLM-5.3-Flash, DeepSeek-V4.1-Flash, Flint, Ripple-65B-Preview)",overall score,49.9,1.32,per solved task,no,https://joelniklaus-harness-leaderboard.hf.space/
|
| 73 |
+
Joel Niklaus Harness Leaderboard,Harness Leaderboard,v0.1.8,2026-11-18,mini-SWE-agent,"open models (GLM-5.3-Flash, DeepSeek-V4.1-Flash, Flint, Ripple-65B-Preview)",overall score,49.9,1.41,per solved task,no,https://joelniklaus-harness-leaderboard.hf.space/
|
| 74 |
+
Joel Niklaus Harness Leaderboard,Harness Leaderboard,v0.1.8,2026-11-18,Pi,"open models (GLM-5.3-Flash, DeepSeek-V4.1-Flash, Flint, Ripple-65B-Preview)",overall score,49.4,1.02,per solved task,no,https://joelniklaus-harness-leaderboard.hf.space/
|
| 75 |
+
Joel Niklaus Harness Leaderboard,Harness Leaderboard,v0.1.8,2026-11-18,Claude Code,"open models (GLM-5.3-Flash, DeepSeek-V4.1-Flash, Flint, Ripple-65B-Preview)",overall score,49.4,1.49,per solved task,no,https://joelniklaus-harness-leaderboard.hf.space/
|
| 76 |
+
Joel Niklaus Harness Leaderboard,Harness Leaderboard,v0.1.8,2026-11-18,Hermes Agent,"open models (GLM-5.3-Flash, DeepSeek-V4.1-Flash, Flint, Ripple-65B-Preview)",overall score,49.3,1.35,per solved task,no,https://joelniklaus-harness-leaderboard.hf.space/
|
| 77 |
+
Joel Niklaus Harness Leaderboard,Harness Leaderboard,v0.1.8,2026-11-18,Codex,"open models (GLM-5.3-Flash, DeepSeek-V4.1-Flash, Flint, Ripple-65B-Preview)",overall score,47.5,1.23,per solved task,no,https://joelniklaus-harness-leaderboard.hf.space/
|
| 78 |
+
Joel Niklaus Harness Leaderboard,Harness Leaderboard,v0.1.8,2026-11-18,Qwen Code,"open models (GLM-5.3-Flash, DeepSeek-V4.1-Flash, Flint, Ripple-65B-Preview)",overall score,43.9,1.33,per solved task,no,https://joelniklaus-harness-leaderboard.hf.space/
|
| 79 |
+
Joel Niklaus Harness Leaderboard,Harness Leaderboard,v0.1.8,2026-11-18,OpenClaw,"open models (GLM-5.3-Flash, DeepSeek-V4.1-Flash, Flint, Ripple-65B-Preview)",overall score,29.7,4.47,per solved task,no,https://joelniklaus-harness-leaderboard.hf.space/
|