JEV Decision Index (0.3)
Community Decision Index for typed decision models (choice, yes/no, and score answers with probabilities), run and published on the Hugging Face Space multimodalart/jev-decision-index by Apolinário (multimodalart). Edition 0.3 (2026-10-06; current point release 0.3.2, 2026-10-12) changes the headline to a Full score: 0.20 × public index + 0.50 × private tests of the same skills + 0.30 × private tasks from new domains, with the private parts equated onto the public scale using 21 untuned stock models. Each benchmark is chance-corrected (0 = random guessing, 100 = perfect).
Models ranked
15
tracked on this benchmark
Score band
65.5 – 4.4
best → lowest tracked
Snapshot trend
—
need ≥2 snapshots
Leaderboard
Tracked models ranked by Decision Index 0.3 (Full score) (higher is better).
Notes: Independent board multimodalart/jev-decision-index, Decision Index 0.3 (point release 0.3.2), entrant 'Perplexity Decider v1.2 (27B)' (engine pplx-decider-v1.2, https://huggingface.co/perplexity-ai/pplx-decider-v1.2). Full score 65.55 (= 0.20 public + 0.50 same-skill private + 0.30 new-domain private, equated). Parts: public index 67.74, same-skill 64.3, new-domain 55.78. Rank 4 by Full score (tie group #2, tied); public rank 3. Board data generated 2026-10-16T19:00:10Z (2026-10-16 12:00 PDT), edition v0.3 dated 2026-10-06, v0.3.2 dated 2026-10-12. Retrieved 2026-10-17; rendered capture evidence/render_board_text.txt. Vendor card claim (68.09 joint public) is NOT what the board shows; board values used.
Notes: Independent board multimodalart/jev-decision-index, Decision Index 0.3 (point release 0.3.2), entrant 'Jev 1.13' (engine jev-1.13.0; board note: 'TypeSafe hosted API, model id jev-1.13.0: the stored rows of the earlier reference run, scored with the same chain as every row'). Full score 60.11 (= 0.20 public + 0.50 same-skill private + 0.30 new-domain private, equated). Parts: public index 57.96, same-skill 58.04, new-domain 54.98. Rank 18 by Full score (tie group #6, tied); public rank 37. Maps to catalog slug jev because that model row names jev-1.13.0 as its current versioned API id (description + versions list). The board's separate Jev 1.14 entry (61.53) is NOT mapped. Board data generated 2026-10-16T19:00:10Z (2026-10-16 12:00 PDT), edition v0.3 dated 2026-10-06, v0.3.2 dated 2026-10-12. Retrieved 2026-10-17; rendered capture evidence/render_board_text.txt.
Notes: Independent board multimodalart/jev-decision-index, Decision Index 0.3 (point release 0.3.2), entrant 'Quyet-1.0-Large' (engine quyet-1.0-large, https://huggingface.co/chinhnc/Quyet-1.0-Large). Full score 58.71 (= 0.20 public + 0.50 same-skill private + 0.30 new-domain private, equated). Parts: public index 60.79, same-skill 58.83, new-domain 49.12. Rank 25 by Full score (tie group #7, tied); public rank 19. Board data generated 2026-10-16T19:00:10Z (2026-10-16 12:00 PDT), edition v0.3 dated 2026-10-06, v0.3.2 dated 2026-10-12. Retrieved 2026-10-17; rendered capture evidence/render_board_text.txt.
Notes: Independent board multimodalart/jev-decision-index, Decision Index 0.3 (point release 0.3.2), entrant 'Surogate Rune 26B-A4B v3' (engine rune-26b-a4b-v3, https://huggingface.co/surogate/rune-26b-a4b-GGUF). Full score 57.43 (= 0.20 public + 0.50 same-skill private + 0.30 new-domain private, equated). Parts: public index 58.28, same-skill 57.09, new-domain 49.38. Rank 35 by Full score (tie group #8, tied); public rank 31. Board data generated 2026-10-16T19:00:10Z (2026-10-16 12:00 PDT), edition v0.3 dated 2026-10-06, v0.3.2 dated 2026-10-12. Retrieved 2026-10-17; rendered capture evidence/render_board_text.txt.
Notes: Independent board multimodalart/jev-decision-index, Decision Index 0.3 (point release 0.3.2), entrant 'Perplexity Decider Flash v1.2 (4B)' (engine pplx-decider-flash-v1.2, https://huggingface.co/perplexity-ai/pplx-decider-flash-v1.2). Full score 54.81 (= 0.20 public + 0.50 same-skill private + 0.30 new-domain private, equated). Parts: public index 60.06, same-skill 56.16, new-domain 43.17. Rank 47 by Full score (tie group #11, tied); public rank 22. Board data generated 2026-10-16T19:00:10Z (2026-10-16 12:00 PDT), edition v0.3 dated 2026-10-06, v0.3.2 dated 2026-10-12. Retrieved 2026-10-17; rendered capture evidence/render_board_text.txt. Vendor card claim (60.05 one-question public) is NOT what the board shows; board values used.
Notes: Independent board multimodalart/jev-decision-index, Decision Index 0.3 (point release 0.3.2), entrant 'Cloudflare clef' (engine clef, https://huggingface.co/Cloudflare/clef). Full score 53.08 (= 0.20 public + 0.50 same-skill private + 0.30 new-domain private, equated). Parts: public index 61.71, same-skill 53.11, new-domain 41.88. Rank 51 by Full score (tie group #12, tied); public rank 17. Board data generated 2026-10-16T19:00:10Z (2026-10-16 12:00 PDT), edition v0.3 dated 2026-10-06, v0.3.2 dated 2026-10-12. Retrieved 2026-10-17; rendered capture evidence/render_board_text.txt.
Notes: Independent board multimodalart/jev-decision-index, Decision Index 0.3 (point release 0.3.2), entrant 'Winnow-12B' (engine winnow-12b-q8, https://huggingface.co/EldanRing/Winnow-12B). Full score 49.27 (= 0.20 public + 0.50 same-skill private + 0.30 new-domain private, equated). Parts: public index 50.66, same-skill 48.14, new-domain 44.16. Rank 61 by Full score (tie group #15, tied); public rank 67. Board data generated 2026-10-16T19:00:10Z (2026-10-16 12:00 PDT), edition v0.3 dated 2026-10-06, v0.3.2 dated 2026-10-12. Retrieved 2026-10-17; rendered capture evidence/render_board_text.txt.
Notes: Independent board multimodalart/jev-decision-index, Decision Index 0.3 (point release 0.3.2), entrant 'Cloudflare clef-omni (30B-A3B)' (engine clef-omni, https://huggingface.co/Cloudflare/clef-omni). Full score 48.28 (= 0.20 public + 0.50 same-skill private + 0.30 new-domain private, equated). Parts: public index 56.54, same-skill 46.51, new-domain 40.84. Rank 66 by Full score (tie group #16, tied); public rank 51. Board data generated 2026-10-16T19:00:10Z (2026-10-16 12:00 PDT), edition v0.3 dated 2026-10-06, v0.3.2 dated 2026-10-12. Retrieved 2026-10-17; rendered capture evidence/render_board_text.txt.
Notes: Independent board multimodalart/jev-decision-index, Decision Index 0.3 (point release 0.3.2), entrant 'Cloudflare clef-flash' (engine clef-flash, https://huggingface.co/Cloudflare/clef-flash). Full score 47.61 (= 0.20 public + 0.50 same-skill private + 0.30 new-domain private, equated). Parts: public index 56.15, same-skill 44.01, new-domain 42.5. Rank 69 by Full score (tie group #17, tied); public rank 52. Board data generated 2026-10-16T19:00:10Z (2026-10-16 12:00 PDT), edition v0.3 dated 2026-10-06, v0.3.2 dated 2026-10-12. Retrieved 2026-10-17; rendered capture evidence/render_board_text.txt.
Notes: Independent board multimodalart/jev-decision-index, Decision Index 0.3 (point release 0.3.2), entrant 'LiquidAI d1-3B' (engine d1-3b, https://huggingface.co/LiquidAI/d1-3B). Full score 38.9 (= 0.20 public + 0.50 same-skill private + 0.30 new-domain private, equated). Parts: public index 48.99, same-skill 41.23, new-domain 27.96. Rank 99 by Full score (tie group #24, tied); public rank 72. Board data generated 2026-10-16T19:00:10Z (2026-10-16 12:00 PDT), edition v0.3 dated 2026-10-06, v0.3.2 dated 2026-10-12. Retrieved 2026-10-17; rendered capture evidence/render_board_text.txt.
Notes: Independent board multimodalart/jev-decision-index, Decision Index 0.3 (point release 0.3.2), entrant 'Interfaze lev Qwen3.5-4B LoRA' (engine lev, https://huggingface.co/interfaze-ai/lev). Full score 36.53 (= 0.20 public + 0.50 same-skill private + 0.30 new-domain private, equated). Parts: public index 39.2, same-skill 37.2, new-domain 31.98. Rank 108 by Full score (tie group #26, tied); public rank 100. Board data generated 2026-10-16T19:00:10Z (2026-10-16 12:00 PDT), edition v0.3 dated 2026-10-06, v0.3.2 dated 2026-10-12. Retrieved 2026-10-17; rendered capture evidence/render_board_text.txt.
Notes: Independent board multimodalart/jev-decision-index, Decision Index 0.3 (point release 0.3.2), entrant 'Strands Decider 2B' (engine strands-decider-2b, https://huggingface.co/StrandsAgents/strands-decider-2B-hobson-v21). Full score 21.34 (= 0.20 public + 0.50 same-skill private + 0.30 new-domain private, equated). Parts: public index 25.41, same-skill 22.43, new-domain 19.7. Rank 135 by Full score (tie group #38, tied); public rank 139. Board data generated 2026-10-16T19:00:10Z (2026-10-16 12:00 PDT), edition v0.3 dated 2026-10-06, v0.3.2 dated 2026-10-12. Retrieved 2026-10-17; rendered capture evidence/render_board_text.txt.
Notes: Independent board multimodalart/jev-decision-index, Decision Index 0.3 (point release 0.3.2), entrant 'GLiNER2.5-Decide' (engine gliner-decide, https://huggingface.co/fastino/GLiNER2.5-Decide). Full score 9.24 (= 0.20 public + 0.50 same-skill private + 0.30 new-domain private, equated). Parts: public index 11.48, same-skill 12.05, new-domain 9.64. Rank 155 by Full score (tie group #46, tied); public rank 160. Board data generated 2026-10-16T19:00:10Z (2026-10-16 12:00 PDT), edition v0.3 dated 2026-10-06, v0.3.2 dated 2026-10-12. Retrieved 2026-10-17; rendered capture evidence/render_board_text.txt.
Notes: Independent board multimodalart/jev-decision-index, Decision Index 0.3 (point release 0.3.2), entrant 'LiquidAI d1-omni-600M' (engine d1-omni-600m, https://huggingface.co/LiquidAI/d1-omni-600M). Full score 9.09 (= 0.20 public + 0.50 same-skill private + 0.30 new-domain private, equated). Parts: public index 17.9, same-skill 12.86, new-domain 5.08. Rank 156 by Full score (tie group #46, tied); public rank 151. Board data generated 2026-10-16T19:00:10Z (2026-10-16 12:00 PDT), edition v0.3 dated 2026-10-06, v0.3.2 dated 2026-10-12. Retrieved 2026-10-17; rendered capture evidence/render_board_text.txt.
Notes: Independent board multimodalart/jev-decision-index, Decision Index 0.3 (point release 0.3.2), entrant 'Laya' (engine laya-1e28ac20, https://huggingface.co/convaiinnovations/laya). Full score 4.43 (= 0.20 public + 0.50 same-skill private + 0.30 new-domain private, equated). Parts: public index 6.19, same-skill 9.08, new-domain 4.05. Rank 164 by Full score (tie group #49, tied); public rank 167. Board data generated 2026-10-16T19:00:10Z (2026-10-16 12:00 PDT), edition v0.3 dated 2026-10-06, v0.3.2 dated 2026-10-12. Retrieved 2026-10-17; rendered capture evidence/render_board_text.txt.
How to read this benchmark
This benchmark scores models where higher is better. Use scores for directional filtering and shortlisting, not universal quality ranking; then validate pricing, context window, provider availability, and fit for your workload.
Trust this score when
- There is a fresh timestamped snapshot (or multiple snapshots) for this benchmark.
- The model list covers the same version family you can actually deploy today.
- Top candidates overlap with your required routing and feature requirements.
Be cautious when
- There is only one benchmark snapshot or the dataset appears stale.
- The benchmark metric direction is opposite of your decision objective.
- The score difference between options is narrow and likely within implementation variance.
Last reviewed: Oct 17, 2026