whatmodel.sh · find the best decision models you can run

Best Decision models for arts & human taste to run locally on a 16GB graphics card

A leaderboard of the best Decision models for arts & human taste you can run locally on a 16GB graphics card (with 32GB of system RAM). Clef-flash is #1 right now, followed by vLLM-SR d3-flash (9B) and Jebadiah 9B v2.1 (T1e). Typed-choice results from the Decision Index 0.4.1 public suite. These are decisions over supplied options, not generated answers. Each model shows the download that fits and the command to run it.

Compare tested decision models and inference recipes across text, vision and workflow tasks. Use task filters and raw results: probability consistency measures coherence, not correctness.

Filter by your hardware and budget → ~16GB fast · local only · reset

ram[2][4][8][16][24][32][48][64][96][128][192][256]GB

gpu[apple silicon][graphics card][unified memory][none]

vram[4][8][12][16][24][32][48][64][80][96][128]GB

price[local][$0.05][$0.1][any]max per M input tokens

Filter by task

task[all tasks][tools & automation][retrieval & classification][language understanding][arts & human taste][knowledge & reasoning][vision][workflows][consistency]

Scores use reviewed arts & human taste evaluations only. Compare scores within this task. Models without task measurements are excluded.

New #1s for your hardware:

orRSS

detecting your OS…

$ terminal

Score vs release date

     │                                                            ● Clef-flash
  60 ┤ · · · · · · · · · · · · · · · · · · · · · · · · · ··· · · ·∘· · · ··· · · · · · · ··· · · · · · · ·∘· ·
     │                                                            │
  40 ┤ · ●─Kev 4B r10─────────────────────────────────────────────∘· · · ··· · · · · · · ··· · · · · · · ··· ·
     │   ∘
     └───┬───────────────┬───────────────┬────────────────┬───────────────┬───────────────┬───────────────┬───
    2026-09-24      2026-09-26      2026-09-28       2026-09-30      2026-10-02      2026-10-04     2026-10-06

Release dates available for 6 of 22 models. Undated models are omitted; month-only dates are plotted at the start of the month.

#	model	task score↓	where↓	RAM↓	speed↓	get it	sources

#	model	task score↓	where↓	RAM↓	get it
44	Clef-flash · bf16 Cloudflare	64±1†	local (slow)	~25GB	how to run ›
65	vLLM-SR d3-flash (9B) · ollaya-f32 vLLM Semantic Router	61±1†	local (slow)	~18GB	how to run ›
75	Jebadiah 9B v2.1 (T1e) · bf16-t1e frontier-infra	58±1†	local (slow)	~25GB	how to run ›
83	GLiClass v4 edge (0.4B) · ollaya-f32 Knowledgator Engineering	57±1†	local	~2.7GB	how to run ›
87	Perplexity Decider Flash v1.2 (4B) · ollaya-f32 Perplexity	56±1†	local	~10GB	how to run ›
102	Bespoke Nimble 9B v3 · ollaya-f32 Bespoke Labs	55±1†	local (slow)	~21GB	how to run ›
115	vLLM-SR d3-mini (4B) · ollaya-f32 vLLM Semantic Router	54±1†	local	~11GB	how to run ›
133	Bespoke Nimble 9B v2 · ollaya-f32 Bespoke Labs	50±1†	local (slow)	~20GB	how to run ›
144	Decider 4B · ollaya-f32 Mapika	47±1†	local	~11GB	how to run ›
157	vLLM-SR Decision 2.0 Lux 9B · ollaya-f32 vLLM Semantic Router	45±1†	local (slow)	~20GB	how to run ›
173	vLLM-SR Decision 2.0 Nox 4B · ollaya-f32 vLLM Semantic Router	40±1†	local	~12GB	how to run ›
179	vLLM-SR d3-lite (0.8B) · ollaya-f32 vLLM Semantic Router	39±1†	local	~4.5GB	how to run ›
188	Kev 4B r10 · ollaya-f32 Jared Palmer	36±1†	local	~11GB	how to run ›
192	vLLM-SR Decision 2.0 Sol 2B · ollaya-f32 vLLM Semantic Router	34±1†	local	~11GB	how to run ›
202	vLLM-SR d3-nano (2B) · ollaya-f32 vLLM Semantic Router	32±1†	local	~8.2GB	how to run ›
205	Strands Decider 2B (Hobson v21) · bf16 Amazon Web Services	31±1	local	~6.6GB	how to run ›
208	vLLM-SR d3-edge (0.6B) · ollaya-f32 vLLM Semantic Router	31±1†	local	~2.6GB	how to run ›
219	vLLM-SR Decision 2.0 Eos 0.8B · ollaya-f32 vLLM Semantic Router	29±1†	local	~5.8GB	how to run ›
235	vLLM-SR Decision 2.0 Kai 0.6B · ollaya-f32 vLLM Semantic Router	28±1†	local	~3.7GB	how to run ›
243	Kev 0.8B r15 · ollaya-f32 Jared Palmer	27±1†	local	~4.6GB	how to run ›
244	LiquidAI d1-omni-600M · ollaya-f32 Liquid AI	26±1†	local	~2.7GB	how to run ›
247	CLM-v0.1-8B Qwen3-8B heads · ollaya-f32 Contrastive-LM	20±1†	local	~15GB	how to run ›

+ 225 more don't fit your machine or budget · show them

247 of 247 catalogue models listed here · sources and licences

* new or thinly tested: ranked on too few benchmark results · † no independently run results verified · ~RAM is estimated; full context may need more memory