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

Best AI models under $0.1 per million tokens

A leaderboard of the best AI models that cost $0.1 or less per million tokens. Hy4 preview (770B-A49B) is #1 right now, followed by MiMo-V2.6-Pro and Qwen3.8 Max (2.4T-A95B). Models are ranked by one score that combines 566 published benchmarks. Filter by your computer or budget to find the best one you can use.

628 models · scores combine 566 benchmarks · updated 2026-10-14 08:02 UTChow it works · find a leaderboard · benchmark your machine

task[all tasks][agent tasks][coding][chat & writing][math][vision & documents][web search]

Filter by your hardware and budget · up to $0.10/M · reset

+ more

set[all sets][LMArena (text) · 1182][ARC Challenge · 8][ANLI Round 1 (prompted) · 8][LAMBADA (OpenAI version), last-word accuracy · 7][ARC Challenge · 3][ARC Challenge · 2]

Filter the model list while keeping overall scores. Models without a shared score scale are not yet calibrated; select a set to see its own scores.

Detect your exact chip or graphics card:

curl -G whatmodel.sh/ --data-urlencode "mac=$(sysctl -n hw.memsize machdep.cpu.brand_string)"Mac

curl -G whatmodel.sh/ --data-urlencode "ram=$(awk '/MemT/{print $2}' /proc/meminfo)" --data-urlencode "nv=$(nvidia-smi --query-gpu=name,memory.total --format=csv,noheader 2>/dev/null)"Linux

curl.exe -G whatmodel.sh/ --data-urlencode "ram=$((gcim Win32_ComputerSystem).TotalPhysicalMemory)" --data-urlencode "nv=$(nvidia-smi --query-gpu=name,memory.total --format=csv,noheader 2>$null)"Windows

$ terminal

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

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

price[local][$0.02][$0.1][$0.5][$2][$10][$50][any]max per M

detecting your OS…

Score vs API price

  80 ┤ · · · · · · MiMo-V2.6-Flash (309B-A15…·●──────●───────────● · · · · · · ∘ · ·∘· · ∘ · ·∘· · · · · ·∘· ·
     │      Qwen3.7 Flash ◆────────●──●───∘∘──┘∘ ∘ ∘  ∘ ∘∘∘∘∘∘ ∘ ∘ ∘∘∘   ∘∘∘∘∘ ∘∘∘ ∘∘∘ · ∘    ∘    ∘      ·
  60 ┤ · ·Granite 4.2 3B ●┘· ·∘∘ ·∘·∘∘∘∘ ∘ ∘ · ·∘·∘· ∘∘∘∘·∘∘ ∘∘· · ∘∘·∘∘∘·∘∘ · ∘ · ∘ · · ·∘· · · · · ∘ · · · ·
  40 ┤ · · · · · · · · · ● ·∘· · · ·∘· ·∘∘ · · · ·∘· ∘ · · · · · · · · · · · · · · · · · · · · · · · ∘ · ··· ·
     │   ●───────────────┘
     └─────────────────────────────────────────────────────────────────────────────────┬───────────┬──────┬───
                                                                                      $2          $4     $6

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

#	model	score↓	where↓	RAM↓	speed↓	context↓	get it	sources
17	Hy4 preview (770B-A49B) Tencent	78±1	local	~248GB		1M	llama.cpp ›
18	MiMo-V2.6-Pro Xiaomi	78±1	local	~443GB		1M	llama.cpp ›
20	Qwen3.8 Max (2.4T-A95B) Alibaba	77±1	local	~380GB		262k	llama.cpp ›	+2
23	DeepSeek V4.1 Flash DeepSeek	76±1	local	~512GB	~12 tok/s	1M	how to run ›
27	Kimi K3 Moonshot AI	76±1	local	~884GB		1M	llama.cpp ›	+6
29	MiMo-V2.6-Flash (309B-A15B) Xiaomi	76±1	local	~136GB		1M	llama.cpp ›
33	DeepSeek V4 Pro (0813) DeepSeek	75±1	local	~1010GB		1M	llama.cpp ›	+2
35	GLM-5.3 Z.ai	75±1	local	~239GB		1M	llama.cpp ›
38	DeepSeek V4 Flash Vision (Exp) DeepSeek	74±2	local	~104GB		1M	llama.cpp ›
40	Qwen3.8-Flash-Next Alibaba	74±1	local	~59GB	~27 tok/s	262k	llama.cpp ›
54	DeepSeek V4 Flash (0731) DeepSeek	73±1	local	~92GB		1M	llama.cpp ›	+2
55	GLM-5.3-Flash Z.ai	73±1	local	~103GB		1M	llama.cpp ›	+2
58	GLM-5.2 Z.ai	73±1	local	~121GB		1M	llama.cpp ›	+6
59	Hunyuan Hy3 (295B-A21B) Tencent	73±1	local	~94GB		256k	llama.cpp ›
61	Qwen3.8 27B Alibaba	72±1	local	~18GB	~6.7 tok/s	262k	LM Studio ›	+3
63	LongCat 2.0 (1.6T-A48B) Meituan	72±2	local	~2155GB		1M	how to run ›
64	Kimi K2.6 Moonshot AI	72±1	local	~317GB		262k	llama.cpp ›	+11
70	DeepSeek V4 Pro (Preview, Apr 2026) DeepSeek	71±1	local	~604GB			llama.cpp ›	+6
72	Kimi K2.7 Code Moonshot AI	71±1	local	~359GB		262k	llama.cpp ›
75	DeepSeek V4 Flash (Preview, Apr 2026) DeepSeek	71±1	local	~103GB			llama.cpp ›	+4
77	Solar Open 2 (250B-A15B) Upstage	71±1†	local	~527GB		1M	how to run ›
80	Ling 3.0 Flash (124B-A5B) inclusionAI (Ant Group)	70±2†	api $0.090/M +local	~71GB		131k	deepinfra.com ↗
83	GLM-5.1 Z.ai	70±0	local	~119GB		200k	llama.cpp ›	+11
85	Qwen3.5 397B-A17B Alibaba	70±0	local	~65GB		262k	llama.cpp ›	+15
87	KAT-Coder-V2.5-Dev (35B-A3B) Kuaishou (Kwaipilot)	70±2†	local	~21GB		262k	LM Studio ›
88	Ornith 1.5 35B-A3B DeepReinforce	70±1†	local	~21GB	~80 tok/s	262k	LM Studio ›
90	MiniMax-M3 MiniMax	69±1	local	~131GB		1M	llama.cpp ›	+2
92	Kimi K2.5 Moonshot AI	69±1	local	~160GB		262k	llama.cpp ›	+15
93	Inkling Small (266B-A12B) Thinking Machines	69±1	local	~86GB		1M	llama.cpp ›
96	GLM-5 Z.ai	69±1	local	~119GB		205k	llama.cpp ›	+11
97	Inkling (975B-A41B) Thinking Machines	69±1	local	~156GB		1M	llama.cpp ›
98	Muse Glimmer 30B Meta	69±1	local	~18GB	~27 tok/s	131k	llama.cpp ›
100	MiMo-V2.5-Pro (1T-A42B) Xiaomi	69±1	local	~163GB		1M	llama.cpp ›
101	Nemotron 3 Ultra 550B-A55B NVIDIA	69±1	local	~89GB		1M	llama.cpp ›
102	Qwen3.6 27B Alibaba	69±1	local	~18GB	~16 tok/s	262k	LM Studio ›	+4
104	A.X K2 (688B-A33B) SK Telecom	69±1†	local	~1454GB		262k	how to run ›
105	MiMo-V2.5 (310B-A15B) Xiaomi	68±1	local	~100GB		1M	llama.cpp ›
109	Qwen3.5 122B-A10B Alibaba	68±1	local	~71GB		262k	LM Studio ›	+5
113	GLM-4.7 Z.ai	68±1	local	~58GB		205k	llama.cpp ›	+5
115	Qwen3.5 27B Alibaba	68±0	local	~18GB	~29 tok/s	262k	LM Studio ›	+8
119	Step 3.5 Flash (196B-A11B) StepFun	68±1	local	~118GB	~18 tok/s	256k	LM Studio ›
121	DeepSeek V3.2 DeepSeek	67±1	local	~110GB		128k	llama.cpp ›	+15
122	Qwen3.6 35B-A3B Alibaba	67±1	local	~21GB	~63 tok/s	262k	LM Studio ›	+4
123	MiniMax-M2.7 MiniMax	67±1	local	~75GB		205k	llama.cpp ›	+2
124	Qwen3-Coder-Next (80B-A3B) Alibaba	67±2	local	~25GB	~62 tok/s	262k	llama.cpp ›
127	Ornith 1.0 35B (35B-A3B) DeepReinforce	67±1	local	~23GB	~103 tok/s	262k	LM Studio ›
130	Hy3 preview (295B-A21B) Tencent	67±1	local	~629GB			how to run ›
131	Gemma 4 31B Google	67±1	local	~24GB	~11 tok/s	262k	LM Studio ›	+20
132	Kimi K2 Thinking Moonshot AI	67±1	local	~160GB		262k	llama.cpp ›
135	K-EXAONE 2.0 (750B-A37B) LG AI Research	67±1†	local	~791GB		262k	how to run ›
137	Qwen3.7 Flash Alibaba	67±1	api $0.055/M			1M	www.alibabacloud.com ↗
138	Laguna M.1 Poolside	66±2	local	~135GB		262k	LM Studio ›
139	DeepSeek V3.1 Terminus DeepSeek	66±1	local	~230GB		164k	llama.cpp ›
140	DeepSeek V3.2 Exp DeepSeek	66±1	local	~726GB			how to run ›
141	GigaChat 3.5 Ultra Reasoning (432B-A28B) Sber	66±1†	local	~461GB		262k	how to run ›
142	Qwen3 235B-A22B Thinking 2507 Alibaba	66±1	local	~76GB		131k	llama.cpp ›	+6
143	Kimi K2 Instruct (0711) Moonshot AI	66±1	local	~152GB		131k	llama.cpp ›	+3
144	Qwen3.5 35B-A3B Alibaba	66±0	local	~21GB	~50 tok/s	262k	LM Studio ›	+12
145	Step 3.7 Flash (198B-A11B) StepFun	66±1	local	~104GB		256k	llama.cpp ›
149	LongCat-Flash-Chat (560B-A27B) Meituan	66±1	local	~333GB		128k	LM Studio ›
150	Spark-X2.5 4B iFLYTEK (SparkLLM)	66±1†	local	~3.5GB	~114 tok/s	1M	LM Studio ›
151	DeepSeek V3.1 DeepSeek	66±1	local	~110GB		131k	llama.cpp ›	+3
153	GLM-4.6 Z.ai	66±1	local	~58GB		205k	llama.cpp ›	+1
156	MiMo-V2-Flash (309B-A15B) Xiaomi	66±1	local	~103GB			llama.cpp ›
157	MiniMax-M2.5 MiniMax	65±1	local	~75GB		205k	llama.cpp ›	+3
160	DeepSeek R1 0528 DeepSeek	65±1	local	~104GB		164k	llama.cpp ›	+6
161	MiniMax-M2.1 MiniMax	65±1	local	~79GB			llama.cpp ›
162	Nemotron Cascade 2 30B-A3B NVIDIA	65±1	local	~20GB	~90 tok/s	256k	LM Studio ›
164	MiMo-V2.6-Distill-Qwen-9B Xiaomi	65±2†	local	~7.8GB			LM Studio ›
165	Laguna XS 2.1 (33B-A3B) Poolside	65±2†	local	~21GB		262k	LM Studio ›
166	K-EXAONE (236B-A23B) LG AI Research	65±1	local	~152GB		262k	llama.cpp ›
167	Qwen3 235B-A22B Instruct 2507 Alibaba	65±1	local	~73GB	~15 tok/s	262k	llama.cpp ›	+3
168	INTELLECT-3 Prime Intellect	65±1	local	~48GB			llama.cpp ›
169	GLM-4.5 Z.ai	65±1	local	~58GB		131k	llama.cpp ›	+3
170	EXAONE 4.5 33B LG AI Research	64±2†	local	~24GB		262k	llama.cpp ›
173	Nemotron 3 Super 120B-A12B NVIDIA	64±1	local	~73GB	~20 tok/s	262k	LM Studio ›
174	MiniMax-M2 MiniMax	64±1	local	~75GB		205k	llama.cpp ›
175	Mistral Medium 3.5 (128B) Mistral	64±1	local	~43GB		262k	llama.cpp ›
176	Laguna XS.2 Poolside	64±1	local	~21GB		262k	LM Studio ›
179	Command A+ (05-2026) Cohere	64±1	local	~71GB		128k	llama.cpp ›
180	Qwen3-VL 235B-A22B Thinking Alibaba	64±1	local	~106GB		256k	LM Studio ›
181	Gemma 4 26B-A4B Google	64±1	local	~18GB	~55 tok/s	262k	LM Studio ›	+12
182	Qwen3-VL 235B-A22B Instruct Alibaba	64±1	local	~128GB		256k	LM Studio ›
183	Kimi K2 Instruct (0905) Moonshot AI	64±2	local	~298GB			llama.cpp ›
184	Nemotron-Labs-3 Puzzle 75B-A9B NVIDIA	64±1†	local	~35GB			llama.cpp ›
185	Qwen3-Coder 480B-A35B Instruct Alibaba	64±1	local	~78GB		262k	llama.cpp ›
188	MiniCPM5 2B OpenBMB	64±1†	local	~2.4GB		131k	LM Studio ›
189	DeepSeek R1 (Jan '25) DeepSeek	64±1	local	~104GB		128k	llama.cpp ›	+15
192	Qwen3-Next 80B-A3B Instruct Alibaba	63±1	local	~26GB	~55 tok/s	131k	llama.cpp ›
194	Qwen3-Next 80B-A3B Thinking Alibaba	63±1	local	~25GB		131k	llama.cpp ›	+3
195	Granite 4.2 30B IBM	63±1	local	~19GB		131k	LM Studio ›
199	GLM-4.5-Air Z.ai	63±1	local	~35GB	~23 tok/s	131k	llama.cpp ›	+3
200	Gemma 4 12B Google	63±1	local	~9.2GB	~29 tok/s	262k	LM Studio ›
201	Trinity-Large-Thinking Arcee AI	63±2	local	~141GB		512k	llama.cpp ›
202	Qwen3.5 9B Alibaba	62±0	local	~6.7GB	~22 tok/s	262k	llama.cpp ›	+17
203	Seed-OSS 36B Instruct ByteDance Seed	62±1	local	~23GB		524k	LM Studio ›
205	MiniMax-M1 (80k) MiniMax	62±1	local	~959GB		1M	how to run ›
206	GLM-4.6V Z.ai	62±2	local	~50GB		128k	llama.cpp ›
209	Nemotron 3.5 Lightning 30B-A3B NVIDIA	62±2	api $0.087/M +local	~20GB	~64 tok/s	262k	fireworks.ai ↗
211	Qwen3-VL 32B Thinking Alibaba	62±1	local	~22GB		256k	LM Studio ›
213	gpt-oss-120b OpenAI	62±1	local	~64GB	~56 tok/s	131k	llama.cpp ›	+19
215	Mistral Large 3 (675B-A41B) Mistral	62±1	local	~108GB		262k	llama.cpp ›	+1
216	Apriel 1.6 15B Thinker ServiceNow	62±1†	local	~10GB			LM Studio ›
218	A.X K1 (519B-A33B) SK Telecom	62±1†	local	~1091GB			how to run ›
219	Devstral 2 (2512) Mistral	62±2	local	~74GB		256k	LM Studio ›
221	GLM-4.5V Z.ai	61±2	local	~69GB		64k	llama.cpp ›
224	Llama 3.1 Nemotron Ultra 253B v1 NVIDIA	61±1	local	~82GB		128k	llama.cpp ›
226	DeepSeek V3 0324 DeepSeek	61±1	local	~102GB		164k	llama.cpp ›	+11
228	Qwen3 30B-A3B Thinking 2507 Alibaba	61±1	local	~19GB		262k	LM Studio ›	+8
230	Qwen3.5 4B Alibaba	61±1	local	~3.8GB		262k	llama.cpp ›	+10
231	Qwen3 30B-A3B Instruct 2507 Alibaba	61±1	local	~19GB	~100 tok/s	262k	LM Studio ›	+4
233	ERNIE 4.5 300B-A47B Baidu	60±1†	local	~97GB		131k	llama.cpp ›
234	Mellum2.1 12B-A2.5B Thinking JetBrains	60±1†	local	~8.3GB			LM Studio ›
236	Nemotron-Cascade-14B-Thinking NVIDIA	60±1†	local	~10GB			LM Studio ›
237	Qwen3-VL 32B Instruct Alibaba	60±1†	local	~22GB		256k	LM Studio ›
238	Devstral Small 2 24B Mistral	60±3†	local	~16GB		262k	llama.cpp ›
239	Gemma 2 9B IT SimPO Princeton NLP	60±1	local	~7.6GB			llama.cpp ›
240	Gemma 2 9B IT DPO (Princeton) Princeton NLP	60±1	local	~7.6GB			llama.cpp ›
241	Falcon-H1R 7B TII	60±1†	local	~5.3GB		262k	LM Studio ›
242	Qwen3 235B-A22B (Apr 2025, hybrid) Alibaba	60±0	local	~73GB		131k	llama.cpp ›	+14
243	DeepSeek V3 (Dec '24) DeepSeek	60±1	local	~337GB		131k	llama.cpp ›	+10
244	Mistral Small 4 (119B-A6.5B) Mistral	60±2	local	~37GB		262k	llama.cpp ›
246	Qwen3-VL 30B-A3B Thinking Alibaba	60±1†	local	~20GB		256k	LM Studio ›
247	Granite 4.2 8B IBM	60±1	local	~6.8GB		131k	LM Studio ›
248	QwQ-32B Alibaba	60±1	local	~21GB		131k	LM Studio ›	+7
249	Solar Open 100B (102B-A12B) Upstage	60±1†	local	~41GB		131k	llama.cpp ›
250	OpenReasoning-Nemotron-32B NVIDIA	59±2†	local	~21GB			LM Studio ›
251	Trinity-Large-Preview Arcee AI	59±2	local	~155GB			llama.cpp ›
252	GLM-4.7-Flash Z.ai	59±1	api $0.0000/M +local	~18GB	~44 tok/s	200k	docs.z.ai ↗
253	EXAONE 4.0 32B LG AI Research	59±1	local	~20GB		131k	LM Studio ›
254	Phi-4-reasoning-plus 14B Microsoft	59±1	local	~10GB			LM Studio ›
256	Qwen3 14B Alibaba	59±1	local	~10GB		33k	LM Studio ›	+6
257	Olmo 3.1 32B Think Ai2	59±1	local	~21GB		66k	LM Studio ›
258	Nemotron-Cascade-8B NVIDIA	59±1†	local	~6.3GB			LM Studio ›
260	gpt-oss-20b OpenAI	58±1	local	~13GB	~79 tok/s	131k	llama.cpp ›	+15
261	Llama 3.3 Nemotron Super 49B v1 NVIDIA	58±2	local	~22GB			llama.cpp ›
262	DeepSeek R1 Distill Llama 70B DeepSeek	58±1	local	~30GB		131k	llama.cpp ›	+1
264	Command A (03-2025) Cohere	58±1	local	~37GB		256k	llama.cpp ›
265	Olmo 3 32B Think Ai2	58±1	local	~21GB		66k	LM Studio ›
266	Phi-4-reasoning 14B Microsoft	58±1	local	~10GB		33k	LM Studio ›
268	Qwen3-VL 30B-A3B Instruct Alibaba	58±1†	local	~20GB		256k	LM Studio ›
269	Qwen3-VL 8B Thinking Alibaba	58±1	local	~6.9GB		256k	llama.cpp ›
271	Magistral Small 1.2 Mistral	58±1	local	~16GB		131k	LM Studio ›
274	Qwen3 32B Alibaba	58±0	local	~20GB		131k	LM Studio ›	+17
275	Nemotron-Cascade-8B-Thinking NVIDIA	58±2†	local	~6.3GB			LM Studio ›
276	Qwen3 Coder 30B-A3B Instruct Alibaba	58±1	local	~19GB	~93 tok/s	160k	LM Studio ›
280	Llama 4 Maverick (402B-A17B) Meta	58±1	local	~65GB		1M	llama.cpp ›	+14
281	Llama 3.3 Nemotron Super 49B v1.5 NVIDIA	58±1	local	~18GB		131k	llama.cpp ›
282	Qwen2.5 Coder 32B Instruct Alibaba	58±1	local	~21GB		131k	LM Studio ›	+5
283	DeepSeek R1 0528 Qwen3 8B DeepSeek	58±1	api $0.068/M +local	~6.5GB	~21 tok/s	128k	novita.ai ↗
284	Granite 4.2 3B IBM	57±1	api $0.052/M +local	~3.3GB		131k	deepinfra.com ↗
286	Nemotron 3 Nano 30B-A3B NVIDIA	57±1	api $0.087/M +local	~20GB	~90 tok/s	262k	deepinfra.com ↗	+3
288	Athene-V2-Chat 72B Nexusflow	57±1	local	~44GB		33k	LM Studio ›
290	Granite 4.1 30B IBM	57±1†	local	~19GB	~25 tok/s		Ollama ›
291	Qwen2.5-Omni 7B Alibaba	57±2	local	~5.8GB			llama.cpp ›
293	Ministral 3 14B Reasoning 2512 Mistral	57±1	local	~10GB		262k	llama.cpp ›
295	Sky-T1 32B Preview NovaSky (UC Berkeley)	57±4	local	~21GB			LM Studio ›
296	Qwen3 4B Thinking 2507 Alibaba	57±1	local	~4.2GB	~67 tok/s	262k	LM Studio ›	+2
297	DeepSeek-V2.5 (1210) DeepSeek	57±1	local	~82GB			llama.cpp ›
298	QED-Nano LM Provers	57±2	local	~4.1GB		262k	llama.cpp ›
299	Mistral Small 3.2 24B Mistral	57±1	local	~15GB		128k	LM Studio ›	+4
301	Qwen3-VL 4B Thinking Alibaba	56±1	local	~4.4GB		256k	llama.cpp ›
302	Qwen2.5-VL 72B Instruct Alibaba	56±1	local	~46GB		33k	LM Studio ›	+3
305	Qwen2.5 72B Instruct Alibaba	56±1	local	~31GB		33k	llama.cpp ›	+24
306	Mistral Large 2 (2411) Mistral	56±1	local	~77GB		128k	llama.cpp ›	+8
307	Llama 3.1 405B Instruct Meta	56±1	local	~160GB		128k	llama.cpp ›	+13
308	Qwen3 8B Alibaba	56±1	local	~6.3GB	~25 tok/s	33k	LM Studio ›	+14
309	DeepSeek R1 Distill Qwen 32B DeepSeek	56±1	local	~21GB	~7.3 tok/s	131k	LM Studio ›	+5
310	Qwen2.5 Coder 14B Instruct Alibaba	56±1	local	~11GB	~18 tok/s	131k	LM Studio ›
311	DeepSeek R1 Distill Qwen 14B DeepSeek	56±1	local	~11GB	~15 tok/s	131k	LM Studio ›	+2
312	Ministral 3 14B Mistral	56±1	local	~9.1GB	~46 tok/s	262k	LM Studio ›	+1
313	Qwen2.5 32B Instruct Alibaba	56±1	local	~21GB		131k	LM Studio ›	+11
314	Nemotron Nano 9B v2 NVIDIA	56±1	local	~7.6GB		131k	llama.cpp ›
315	Molmo 2 8B Ai2	56±2	local	~6.8GB			llama.cpp ›
317	EXAONE Deep 32B LG AI Research	56±2†	local	~22GB		33k	Ollama ›
318	Qwen3 4B Instruct 2507 Alibaba	56±1	local	~4.2GB	~107 tok/s	262k	LM Studio ›	+4
319	Qwen3 30B-A3B (Apr 2025, hybrid) Alibaba	56±0	local	~18GB		33k	LM Studio ›	+10
320	Magistral Small 1.0 (2506) Mistral	56±1	local	~16GB		128k	LM Studio ›
321	Qwen3-VL 8B Instruct Alibaba	56±1	local	~6.9GB		256k	llama.cpp ›
322	Bielik 11B v3.0 Instruct SpeakLeash	56±2	local	~8.1GB			LM Studio ›
323	Spark-X2.5 1.7B iFLYTEK (SparkLLM)	56±1†	local	~1.9GB	~166 tok/s	1M	llama.cpp ›
325	Granite 4.1 8B IBM	55±1	local	~7.1GB	~70 tok/s	131k	LM Studio ›
326	Llama 3.3 70B Instruct Meta	55±1	local	~25GB	~13 tok/s	128k	llama.cpp ›	+21
328	Mistral Large 2 (2407) Mistral	55±1	local	~45GB		128k	llama.cpp ›	+10
329	Falcon-H1 34B Instruct TII	55±3	local	~23GB		262k	llama.cpp ›
330	Qwen2.5-Math 72B Instruct Alibaba	55±2†	local	~44GB		4k	LM Studio ›
331	Llama 4 Scout (109B-A17B) Meta	55±1	local	~36GB	~19 tok/s	328k	llama.cpp ›	+8
332	Bielik 11B v2.6 Instruct SpeakLeash	55±3	local	~8.1GB			LM Studio ›
333	GigaChat 3.5 Ultra Instruct (432B-A28B) Sber	55±1†	local	~457GB		262k	how to run ›
335	Mistral Small 3.1 24B (2503) Mistral	55±1	local	~16GB		128k	llama.cpp ›	+7
336	Llama 3.1 Tulu 3 70B Ai2	55±1	local	~43GB		131k	LM Studio ›
337	Pixtral Large (2411) Mistral	55±3	local	~51GB			llama.cpp ›
338	Olmo 3.1 32B Instruct Ai2	55±2	local	~21GB		66k	LM Studio ›
339	OpenThinker2 32B Open Thoughts	55±1	local	~21GB		33k	LM Studio ›
341	DeepSeek-V2.5 DeepSeek	55±1	local	~151GB		164k	Ollama ›	+1
342	OpenThinker 32B Open Thoughts	55±1†	local	~21GB		33k	LM Studio ›
344	DeepSeek-V2 Chat (0628) DeepSeek	55±1	local	~66GB			llama.cpp ›
345	Command A Reasoning (08-2025) Cohere	55±2	local	~34GB			llama.cpp ›
346	Magistral Small 1.1 (2507) Mistral	55±1†	local	~16GB		40k	LM Studio ›
347	Mellum2 12B-A2.5B Thinking JetBrains	55±1†	local	~8.6GB		131k	LM Studio ›
348	Ministral 3 8B Reasoning 2512 Mistral	55±1	local	~6.5GB		262k	LM Studio ›
349	Nemotron 3 Nano 4B NVIDIA	54±1	local	~2.9GB	~52 tok/s	262k	LM Studio ›
350	Llama 3.1 Tulu 3 70B DPO Ai2	54±1	local	~22GB		131k	llama.cpp ›
354	DeepSeek-Coder-V2 Instruct (0614) DeepSeek	54±1	local	~92GB			llama.cpp ›	+2
355	DeepSeek-Coder-V2 Instruct (0724) DeepSeek	54±1	local	~151GB			llama.cpp ›
356	EXAONE Deep 7.8B LG AI Research	54±2†	local	~6.5GB		33k	llama.cpp ›
357	Mistral Small 3 24B (2501) Mistral	54±1	local	~15GB		32k	LM Studio ›	+6
360	GLM-4 9B Chat 1M Zhipu AI (THUDM)	54±4	local	~8.1GB		1M	llama.cpp ›
363	Qwen2-VL 72B Instruct Alibaba	54±1	local	~34GB			llama.cpp ›
364	Qwen2.5-VL 32B Instruct Alibaba	54±2	local	~23GB		33k	LM Studio ›	+1
366	Llama 3.1 70B Instruct Meta	54±0	local	~27GB		128k	llama.cpp ›	+22
367	Gemma 3 27B Google	54±1	api $0.10/M +local	~19GB		131k	deepinfra.com ↗	+23
368	Hermes 3 Llama 3.1 405B Nous Research	54±1	local	~114GB		131k	llama.cpp ›
370	Phi-4 14B Microsoft	53±1	api $0.087/M +local	~10GB		16k	deepinfra.com ↗	+11
372	Bielik 11B v2.1 Instruct SpeakLeash	53±3	local	~8.5GB			llama.cpp ›
373	Ministral 3 8B Mistral	53±1	local	~6.5GB	~68 tok/s	262k	LM Studio ›	+2
374	Olmo 3 7B Think Ai2	53±1	local	~5.8GB		66k	LM Studio ›
375	s1.1 32B Simple Scaling	53±1	local	~22GB		33k	LM Studio ›
376	Qwen3-VL 4B Instruct Alibaba	53±1	local	~4.4GB		256k	llama.cpp ›
377	Gemma 4 E4B Google	53±1	local	~6.4GB	~21 tok/s	131k	LM Studio ›	+7
378	Qwen2.5-Math 7B Instruct Alibaba	53±2†	local	~5.4GB		4k	LM Studio ›
379	Llama 3.1 Nemotron 70B Instruct NVIDIA	53±1	local	~43GB		128k	LM Studio ›	+4
380	LFM2.5 2.6B Liquid AI	53±1	local	~2.3GB		131k	LM Studio ›
382	Bielik 11B v2.2 Instruct SpeakLeash	53±3	local	~8.5GB			llama.cpp ›
386	Granite 4.0 H Small (32B-A9B) IBM	53±1	local	~21GB		131k	LM Studio ›	+1
387	LFM2.5 8B-A1B Liquid AI	53±1	local	~5.5GB	~174 tok/s	128k	LM Studio ›
388	Llama 3.1 Nemotron 51B Instruct NVIDIA	53±2	local	~34GB			llama.cpp ›
389	Aya Expanse 32B Cohere	53±2	local	~22GB		128k	llama.cpp ›	+1
390	Qwen2.5 Coder 7B Instruct Alibaba	53±1	local	~5.4GB	~3.8 tok/s	131k	LM Studio ›
391	OpenCoder-8B-Instruct INF (infly)	53±1	local	~6.4GB			llama.cpp ›
392	Gemma 3 12B Google	53±1	api $0.075/M +local	~10GB	~13 tok/s	131k	deepinfra.com ↗	+10
393	Qwen2.5 14B Instruct Alibaba	53±1	local	~11GB		131k	LM Studio ›	+7
395	Cogito v1 Preview Llama 8B Deep Cogito	52±4	local	~6.2GB		128k	LM Studio ›
396	Llama-3-Instruct-8B-SimPO Princeton NLP	52±1	local	~6.6GB			llama.cpp ›
397	Llama-3-Instruct-8B-SimPO-ExPO Chujie Zheng (ExPO)	52±1	local	~6.6GB			llama.cpp ›
398	Athene-70B Nexusflow	52±1	local	~29GB			llama.cpp ›
399	Jamba 1.5 Large AI21 Labs	52±1	local	~839GB			how to run ›
400	Qwen2 72B Instruct Alibaba	52±1	local	~33GB		131k	llama.cpp ›	+14
402	CodeQwen1.5 7B Chat Alibaba	52±1	local	~6.4GB		64k	llama.cpp ›
403	OpenCodeInterpreter-DS-33B M-A-P	52±1	local	~22GB			LM Studio ›
404	LFM2 24B-A2B Liquid AI	52±1	local	~14GB		33k	LM Studio ›
406	Qwen3 4B Alibaba	52±1	local	~4GB		33k	LM Studio ›	+8
407	Bielik 11B v2.0 Instruct SpeakLeash	52±3	local	~8.5GB			llama.cpp ›
409	DeepSeek-Coder 33B Instruct DeepSeek	51±1	local	~21GB		16k	LM Studio ›	+1
410	Codestral 22B v0.1 Mistral AI	51±1	local	~15GB		32k	LM Studio ›
412	Gemma 2 27B IT Google	51±1	local	~19GB		8k	LM Studio ›	+10
413	Llama 3 70B Instruct Meta	51±0	local	~24GB		8k	llama.cpp ›	+17
414	Llama 3.2 90B Vision Instruct Meta	51±1	local	~58GB		128k	Ollama ›	+1
415	Bielik 11B v2.3 Instruct SpeakLeash	51±2	local	~8.6GB			llama.cpp ›
416	WizardCoder 33B V1.1 WizardLM	51±1	local	~22GB		16k	llama.cpp ›
418	Llama-3-8B-ProLong-512k-Instruct Princeton NLP	51±4	local	~35GB			how to run ›
420	Mellum2 12B-A2.5B Instruct JetBrains	51±2†	local	~7.6GB		131k	LM Studio ›
421	WizardLM-2 8x22B Microsoft	51±1	local	~91GB			Ollama ›
422	Mistral Small Instruct 22B (2409) Mistral AI	51±2	local	~15GB		32k	LM Studio ›
423	OpenThinker3 7B Open Thoughts	51±1	local	~6GB		33k	LM Studio ›
424	Granite 4.1 3B IBM	51±1†	local	~3.7GB	~131 tok/s		Ollama ›
425	OpenCodeInterpreter-DS-6.7B M-A-P	51±1	local	~5.8GB			llama.cpp ›
426	Qwen2-Math 72B Instruct Alibaba	51±1	local	~32GB		4k	llama.cpp ›
427	Molmo 72B (0924) Ai2	51±2	local	~309GB			how to run ›
428	AceReason-Nemotron-1.0-14B NVIDIA	51±2†	local	~10GB			LM Studio ›
429	Artigenz-Coder-DS-6.7B Artigenz	51±1	local	~5.8GB			llama.cpp ›
430	Yi 34B 200K (base) 01.AI	51±4	local	~23GB		200k	llama.cpp ›
431	Hermes 3 Llama 3.1 70B Nous Research	51±1	local	~43GB		131k	Ollama ›
432	OLMo 2 32B Instruct (0325) Ai2	50±1	local	~21GB			LM Studio ›
433	speechless-codellama-34b-v2.0 uukuguy	50±1	local	~23GB			llama.cpp ›
434	Ministral 3 3B Reasoning 2512 Mistral	50±1	local	~3.4GB		262k	LM Studio ›
435	Aya 23 35B Cohere For AI	50±2	local	~22GB		8k	LM Studio ›
436	Qwen2.5-VL 7B Instruct Alibaba	50±1	local	~5.8GB			llama.cpp ›
438	Mixtral 8x22B Instruct v0.1 Mistral	50±1	local	~56GB		64k	llama.cpp ›	+9
439	EXAONE Deep 2.4B LG AI Research	50±2†	local	~3.2GB		33k	Ollama ›
440	Magicoder-S-DS 6.7B UIUC (ise-uiuc)	50±1	local	~5.8GB			llama.cpp ›
441	Pixtral 12B (2409) Mistral	50±2	local	~9.6GB			llama.cpp ›
443	Qwen2.5 7B Instruct Alibaba	50±1	local	~5.4GB	~54 tok/s	131k	LM Studio ›	+12
444	Yi-Coder 9B Chat 01.AI	50±1	local	~6.3GB		128k	LM Studio ›
445	InternLM2.5 7B Chat Shanghai AI Laboratory (InternLM)	50±2	local	~6.1GB			LM Studio ›
446	XwinCoder-34B Xwin-LM	50±1	local	~23GB			llama.cpp ›
447	Code Millenials 34B Bud Ecosystem	50±1	local	~23GB			llama.cpp ›
448	LFM2.5 1.2B Thinking Liquid AI	50±2	local	~1.3GB		33k	LM Studio ›
449	StarChat2 15B v0.1 Hugging Face	50±1	local	~12GB		16k	llama.cpp ›
450	WaveCoder-Ultra-6.7B Microsoft	50±1	local	~5.8GB			llama.cpp ›
452	Qwen2-Math 7B Instruct Alibaba	50±2†	local	~6GB		4k	LM Studio ›
455	Qwen3-VL 2B Thinking Alibaba	50±1	local	~2.7GB		256k	llama.cpp ›
456	Phi-3.5-MoE-instruct Microsoft	49±1	local	~25GB		128k	LM Studio ›
457	MegaBeam-Mistral-7B-512k AWS (aws-prototyping)	49±4	local	~5.9GB			llama.cpp ›
458	Gemma 3n E4B Google	49±1	local	~5.6GB		32k	llama.cpp ›
460	OpenThinker2 7B Open Thoughts	49±2†	local	~6.4GB		33k	llama.cpp ›
461	DeepSeek-Coder 7B Instruct v1.5 DeepSeek	49±1	local	~5.9GB			llama.cpp ›
462	Hermes 2 Theta Llama-3 70B Nous Research	49±2	local	~47GB		8k	llama.cpp ›
463	DeepSeek-Coder 6.7B Instruct DeepSeek	49±1	local	~8.8GB		16k	llama.cpp ›
464	Qwen2 57B-A14B Instruct Alibaba	49±2	local	~35GB		33k	LM Studio ›
465	Granite 4.0 H Tiny (7B-A1B) IBM	49±1	local	~4.5GB	~12 tok/s	131k	LM Studio ›
467	MiniCPM3-4B OpenBMB	49±2	local	~4GB		32k	LM Studio ›
468	Bielik 4.5B v3.0 Instruct SpeakLeash	49±2	local	~4.3GB			LM Studio ›
469	Qwen2.5 Coder 3B Instruct Alibaba	49±1	local	~2.6GB	~9.5 tok/s	33k	LM Studio ›
470	Gemma 2 9B IT Google	49±1	local	~8.9GB		8k	LM Studio ›	+14
471	Yi-1.5 34B (base) 01.AI	49±2	local	~22GB		4k	LM Studio ›
473	Command R Plus (08-2024) Cohere	49±1	local	~63GB		128k	LM Studio ›	+1
474	GLM-4 9B Chat Zhipu AI (Z.ai)	49±2	local	~7.1GB		128k	Ollama ›	+1
475	AceReason-Nemotron-1.1-7B NVIDIA	49±2†	local	~6GB			LM Studio ›
476	WhiteRabbitNeo-33B-v1 WhiteRabbitNeo	49±1	local	~22GB			llama.cpp ›
477	Qwen1.5 110B Chat Alibaba	49±1	local	~45GB		32k	llama.cpp ›	+1
478	speechless-coder-ds-6.7b uukuguy	48±1	local	~5.8GB			llama.cpp ›
479	Magicoder-S-CL-7B UIUC (ise-uiuc)	48±1	local	~5.8GB			llama.cpp ›
480	Phind CodeLlama 34B v2 Phind	48±1	local	~23GB		16k	Ollama ›
481	DeepSeek R1 Distill Qwen 7B DeepSeek	48±1	local	~5.4GB	~35 tok/s	131k	LM Studio ›	+6
482	Jamba 1.5 Mini AI21 Labs	48±1	local	~34GB			llama.cpp ›
484	InternVL2-26B Shanghai AI Laboratory (OpenGVLab)	48±2	local	~55GB			how to run ›
485	Aya Expanse 8B Cohere	48±2	local	~6.4GB		8k	LM Studio ›
488	EuroLLM 9B Instruct UTTER (EuroLLM)	48±2	local	~6.9GB		4k	LM Studio ›
491	Granite 4.0 H Micro 3B IBM	48±1	local	~3.4GB		131k	LM Studio ›
493	Code Llama 70B Instruct Meta	48±1	local	~28GB		16k	llama.cpp ›	+1
494	OLMo 2 13B Instruct (1124) Allen Institute for AI	48±1	local	~9.6GB		4k	LM Studio ›
495	InternLM2.5 7B Chat 1M Shanghai AI Laboratory (InternLM)	48±4	local	~6.4GB			llama.cpp ›
496	Falcon-H1 7B Instruct TII	48±3	local	~6GB		262k	LM Studio ›
497	Qwen3.5 2B Alibaba	48±1	local	~2.2GB	~137 tok/s	262k	llama.cpp ›	+7
500	Llama 3 70B (base) Meta	48±2	local	~43GB			LM Studio ›
501	MedGemma 4B IT Google	48±3	local	~4.1GB			llama.cpp ›
502	Nova Micro Amazon	47±2	api $0.061/M			128k	docs.aws.amazon.com ↗
503	Gemma 4 E2B Google	47±1	api $0.050/M +local	~4.5GB		131k	docs.aws.amazon.com ↗	+6
504	OpenChat 3.5 0106 OpenChat	47±1	local	~6.1GB		8k	llama.cpp ›	+1
505	Command R (08-2024) Cohere	47±1	local	~21GB		128k	LM Studio ›	+3
506	Tulu 2 70B Allen Institute for AI	47±2†	local	~28GB			llama.cpp ›
507	Hermes 2 Pro Llama 3 8B Nous Research	47±3	local	~6.6GB		8k	llama.cpp ›
508	EXAONE 4.0 1.2B LG AI Research	47±1†	local	~2.2GB		66k	LM Studio ›
511	Yi 1.5 34B Chat 01.AI	47±1	local	~22GB		4k	LM Studio ›	+5
512	DBRX Instruct Databricks	47±1	local	~85GB		32k	Ollama ›	+3
515	Granite 34B Code Instruct (8K) IBM	47±1	local	~22GB			Ollama ›
516	WizardCoder Python 34B V1.0 WizardLM	47±1	local	~22GB		16k	Ollama ›
518	Granite 4.0 Micro 3B IBM	47±1	local	~3.5GB	~6.2 tok/s	131k	LM Studio ›
519	Cogito v1 Preview Llama 3B Deep Cogito	47±4	local	~3.4GB		128k	LM Studio ›
520	InternLM2.5 20B Chat Shanghai AI Laboratory (InternLM)	47±2	local	~14GB		33k	llama.cpp ›
521	StarCoder2-15B-Instruct-v0.1 BigCode	47±1	local	~11GB		16k	Ollama ›
522	Qwen2.5-Math 1.5B Instruct Alibaba	47±2†	local	~1.8GB		4k	LM Studio ›
524	Qwen1.5 72B Chat Alibaba	47±1	local	~31GB		32k	llama.cpp ›	+9
525	Granite 3.3 8B Instruct IBM	47±1	local	~6.3GB		128k	LM Studio ›
526	speechless-starcoder2-15b uukuguy	47±1	local	~12GB			llama.cpp ›
527	OpenChat 3.5 0106 Gemma OpenChat	47±2	local	~6.9GB		8k	llama.cpp ›
528	Qwen2-VL 7B Instruct Alibaba	47±1	local	~5.8GB			LM Studio ›
529	DeepSeek R1 Distill Llama 8B DeepSeek	47±1	local	~6.2GB		131k	LM Studio ›	+1
531	Phi-3 Medium 128K Instruct Microsoft	47±1	local	~10GB		128k	LM Studio ›
532	Zephyr ORPO 141B-A35B v0.1 Hugging Face	47±2	local	~56GB		66k	llama.cpp ›
535	Phi-4-mini-reasoning 3.8B Microsoft	47±2	local	~4GB		128k	LM Studio ›
536	Phi-3-medium-4k-instruct Microsoft	47±1	local	~10GB		4k	LM Studio ›	+1
537	QwQ-32B-Preview Alibaba	46±2	local	~21GB		33k	LM Studio ›	+2
538	DeepSeek-Coder-V2 Lite Instruct (16B-A2.4B) DeepSeek	46±1	local	~9.6GB		128k	LM Studio ›	+1
541	Llama 3.1 8B Instruct Meta	46±0	local	~6.2GB	~47 tok/s	128k	LM Studio ›	+25
542	Ministral 3 3B Mistral	46±1	api $0.10/M +local	~3.4GB	~128 tok/s	262k	docs.mistral.ai ↗	+2
543	Olmo 3 7B Instruct Ai2	46±1	local	~5.8GB		66k	LM Studio ›
544	Qwen2-Math 1.5B Instruct Alibaba	46±2†	local	~2.4GB		4k	LM Studio ›
545	Command R Plus (04-2024) Cohere	46±1	local	~43GB		128k	llama.cpp ›	+6
547	Mistral NeMo Instruct (2407) Mistral	46±1	local	~8.8GB		128k	LM Studio ›	+9
549	Yi 1.5 9B Chat 01.AI	46±1	local	~6.8GB		4k	LM Studio ›	+1
550	Llama 3.1 Tulu 3 8B Ai2	46±1	local	~6.2GB		131k	LM Studio ›
551	Phi-3-mini-4k-instruct (June 2024 update) Microsoft	46±1	local	~5.7GB			LM Studio ›
552	Qwen1.5 32B Chat Alibaba	46±1	local	~22GB		32k	llama.cpp ›
553	Xwin-LM 70B v0.1 Xwin-LM	46±1	local	~28GB			llama.cpp ›
554	Eurus-2-7B-PRIME PRIME-RL	46±2	local	~6.4GB		4k	llama.cpp ›
555	Qwen2 7B Instruct Alibaba	46±1	local	~5.4GB		131k	LM Studio ›	+4
556	Gemma 3 4B Google	46±1	api $0.063/M +local	~3.8GB	~74 tok/s	131k	deepinfra.com ↗	+11
557	LLaVA-OneVision Qwen2 72B (ov-sft) LLaVA (lmms-lab)	46±3	local	~155GB			how to run ›
558	Granite 20B Code Instruct (8K) IBM	46±1	local	~14GB			Ollama ›
559	Command R7B (12-2024) Cohere	46±1	local	~6.2GB		128k	LM Studio ›
560	Dolphin 2.6 Mixtral 8x7B Cognitive Computations	46±1	local	~29GB			llama.cpp ›
561	Phi-4-mini 3.8B Microsoft	45±1	local	~4GB		128k	LM Studio ›	+4
562	Ministral 8B (2410) Mistral	45±1	local	~6.2GB		128k	LM Studio ›	+6
563	Granite 3.0 8B Instruct IBM	45±1	local	~6.6GB		4k	llama.cpp ›
564	Granite 8B Code Instruct (4K) IBM	45±1	local	~6.6GB			llama.cpp ›
565	Falcon3 7B Instruct TII	45±3	local	~5.9GB		32k	LM Studio ›
566	Code Llama 70B Python Meta	45±1	local	~28GB		16k	llama.cpp ›
567	Code Tulu 2 34B Allen Institute for AI	45±2†	local	~72GB			how to run ›
568	OpenThinker 7B Open Thoughts	45±2†	local	~6GB		33k	LM Studio ›
569	Falcon3 10B Instruct TII	45±3	local	~7.6GB	~13 tok/s	32k	LM Studio ›
570	Granite 3.2 8B Instruct IBM	45±1	local	~6.6GB		131k	llama.cpp ›
571	Yi 34B Chat 01.AI	45±1	local	~23GB		4k	llama.cpp ›	+1
572	Hermes 2 Theta Llama 3 8B Nous Research	45±2	local	~6.6GB		8k	llama.cpp ›
573	Phi-3.5-mini-instruct Microsoft	45±1	local	~5.7GB		128k	LM Studio ›	+4
574	Qwen2.5-VL 3B Instruct Alibaba	45±1	local	~2.8GB			llama.cpp ›
575	Llama 3.1 8B (base) Meta	45±3	local	~6.2GB			LM Studio ›
576	Gemma 3n E2B IT Google	45±2	local	~4.1GB		32k	llama.cpp ›
577	Code Llama 70B (base) Meta	45±1	local	~28GB		16k	llama.cpp ›
578	Qwen3-VL 2B Instruct Alibaba	45±1	local	~2.7GB		256k	llama.cpp ›
581	LFM2.5 1.2B Instruct Liquid AI	45±2	local	~1.3GB		33k	LM Studio ›
582	Qwen3 1.7B Alibaba	45±1	local	~2.6GB	~96 tok/s	33k	LM Studio ›	+10
583	Granite 3.1 8B Instruct IBM	45±1	local	~6.7GB		128k	llama.cpp ›
584	Llama 3.2 11B Vision Instruct Meta	45±2	local	~8.9GB		128k	Ollama ›
587	CogVLM2 Llama3 Chat 19B Zhipu AI (THUDM)	44±3	local	~43GB			how to run ›
588	Qwen2.5 3B Instruct Alibaba	44±1	local	~2.6GB	~61 tok/s	33k	LM Studio ›	+5
589	LFM2 2.6B Liquid AI	44±1	local	~2GB		33k	LM Studio ›
590	Qwen2-Audio 7B Instruct Alibaba	44±3	local	~6.5GB			llama.cpp ›
591	MiniCPM-V 2.6 OpenBMB	44±2	local	~5.8GB			llama.cpp ›
592	Tulu 2 DPO 70B Ai2	44±1	local	~32GB			llama.cpp ›	+1
593	Phi-3-small-8k-instruct Microsoft	44±2	local	~17GB			how to run ›
594	CodeGemma 7B IT (v1.0) Google	44±1	local	~7.1GB			llama.cpp ›
595	Yi-1.5 9B (base) 01.AI	44±2	local	~6.8GB		4k	LM Studio ›
597	DeepSeek-Coder 1.3B Instruct DeepSeek	44±1	local	~2.9GB		16k	Ollama ›
598	Aya 23 8B Cohere For AI	44±3	local	~6.2GB		8k	LM Studio ›
599	Qwen2.5 Coder 1.5B Instruct Alibaba	44±1	local	~1.8GB		33k	LM Studio ›
600	SmolLM3 3B Hugging Face	44±1	local	~2.9GB		66k	LM Studio ›
601	WizardLM 70B V1.0 Microsoft	44±1	local	~32GB			llama.cpp ›
602	Mathstral 7B v0.1 Mistral AI	44±1	local	~5.6GB		32k	LM Studio ›
603	Aya Vision 32B Cohere	44±3	local	~71GB		16k	how to run ›
604	Hermes 3 Llama 3.1 8B Nous Research	44±1	local	~6.6GB		128k	llama.cpp ›
605	Llama 3 8B Instruct Meta	44±0	local	~6.6GB		8k	llama.cpp ›	+19
606	WizardCoder Python 13B V1.0 WizardLM	44±2	local	~15GB		16k	llama.cpp ›
607	OpenChat 3.5 OpenChat	43±1	local	~6.1GB		8k	llama.cpp ›
608	Command R (03-2024) Cohere	43±1	local	~33GB		128k	llama.cpp ›	+6
609	Granite 34B Code Base (8K) IBM	43±1	local	~24GB			llama.cpp ›
610	speechless-starcoder2-7b uukuguy	43±1	local	~6.1GB			llama.cpp ›
611	Falcon2 11B (base) TII	43±3	local	~8.2GB		8k	Ollama ›
612	Snowflake Arctic Instruct Snowflake	43±2	local	~1007GB			how to run ›
613	OLMo 2 7B Instruct (1124) Allen Institute for AI	43±1	local	~5.8GB		4k	LM Studio ›
614	WizardCoder-15B-V1.0 WizardLM	43±1	local	~12GB			llama.cpp ›
616	Phi-3-vision-128k-instruct Microsoft	43±2	local	~10GB			how to run ›
618	DeepSeek LLM 67B Chat DeepSeek	43±1	local	~31GB		4k	llama.cpp ›
619	Qwen1.5 14B Chat Alibaba	43±1	local	~11GB		32k	Ollama ›	+3
620	Llama 2 70B (base) Meta	43±1	local	~32GB		4k	llama.cpp ›
621	Mixtral 8x7B v0.1 (base) Mistral AI	43±1	local	~29GB		32k	llama.cpp ›
622	InternVL2-4B Shanghai AI Laboratory (OpenGVLab)	43±2	local	~10GB			how to run ›
623	Code Llama 34B Python Meta	43±1	local	~23GB		16k	Ollama ›
624	DeepSeek-Coder 33B Base DeepSeek	43±1	local	~22GB		16k	llama.cpp ›
625	speechless-coding-7b-16k-tora uukuguy	43±1	local	~16GB			how to run ›
626	InternLM2 Chat 20B Shanghai AI Laboratory (InternLM)	43±1	local	~14GB			llama.cpp ›
627	CodeQwen1.5 7B (base) Alibaba	43±1	local	~6.4GB		66k	llama.cpp ›
628	Mixtral 8x7B Instruct v0.1 Mistral	42±0	local	~28GB		32k	LM Studio ›	+15
629	Nous Hermes 2 Mistral 7B DPO Nous Research	42±3	local	~6.1GB		33k	llama.cpp ›
631	Granite 4.0 1B IBM	42±1	local	~2.4GB		128k	LM Studio ›
632	OpenHermes-2.5-Code-290k-13B ajibawa-2023	42±1	local	~9.8GB			llama.cpp ›
633	Granite 4.0 H 1B IBM	42±1	local	~2.3GB		131k	LM Studio ›
634	Nous Hermes 2 Mixtral 8x7B DPO Nous Research	42±2	local	~30GB		33k	llama.cpp ›
635	Molmo 7B-D (0924) Ai2	42±1	local	~35GB			how to run ›
636	North Micro Vision Instruct (2.4B) Cohere	42±2†	local	~3.8GB		128k	how to run ›
637	Deita 7B v1.0 HKUST NLP	42±2	local	~32GB			how to run ›
639	OpenHermes 2.5 Mistral 7B Nous Research	42±1	local	~5.9GB			llama.cpp ›	+2
640	Code Llama 34B Instruct Meta	42±1	local	~22GB		16k	LM Studio ›	+1
641	Starling-LM 7B beta Nexusflow	42±1	local	~6GB		8k	LM Studio ›	+3
642	Code Llama 34B (base) Meta	42±1	local	~23GB		16k	Ollama ›	+1
643	Vicuna 33B v1.3 LMSYS	41±1	local	~23GB		2k	Ollama ›	+2
644	Phi-3-mini-128k-instruct (Apr 2024) Microsoft	41±2	local	~9.5GB			how to run ›	+1
645	Bielik 1.5B v3.0 Instruct SpeakLeash	41±3	local	~2.4GB			LM Studio ›
646	Starling-LM 7B alpha UC Berkeley (NEST)	41±1	local	~6.1GB		8k	llama.cpp ›	+2
647	LLaVA-v1.6-34B LLaVA	41±2	local	~24GB		4k	Ollama ›
648	Code Tulu 2 13B Allen Institute for AI	41±2†	local	~29GB			how to run ›
649	Llama3-OpenBioLLM-8B Saama AI Labs	41±2	local	~6.6GB			llama.cpp ›
650	Mistral 7B Instruct v0.3 Mistral	41±1	local	~5.6GB		32k	LM Studio ›	+4
651	Granite 3.3 2B Instruct IBM	41±1†	local	~3GB		128k	LM Studio ›
652	DeepSeek-Coder 6.7B Base DeepSeek	41±1	local	~8.8GB		16k	llama.cpp ›
653	Granite 3.0 2B Instruct IBM	41±1	local	~3.2GB		4k	llama.cpp ›
654	LFM2 1.2B Liquid AI	41±1	local	~1.3GB		33k	LM Studio ›
655	Llama 2 70B Chat Meta	41±1	local	~32GB		4k	llama.cpp ›	+13
656	Llama 3.2 3B Instruct Meta	41±1	local	~3.3GB	~139 tok/s	131k	LM Studio ›	+11
657	Granite 8B Code Base (4K) IBM	41±1	local	~6.6GB			llama.cpp ›
658	StarCoder2 15B BigCode	41±1	local	~11GB		16k	LM Studio ›
659	Falcon-H1 3B Instruct TII	41±3	local	~3.4GB		131k	LM Studio ›
660	Phi-2 (2.7B) Microsoft	40±1	local	~4.7GB		2k	LM Studio ›	+2
661	speechless-code-mistral-7b-v1.0 uukuguy	40±1	local	~17GB			how to run ›
662	Llama-3-Instruct-8B-SimPO-v0.2 Princeton NLP	40±1	local	~6.6GB			llama.cpp ›
663	Qwen-14B-Chat Alibaba	40±1	local	~11GB			llama.cpp ›
664	Tulu 2 13B Allen Institute for AI	40±2	local	~9.8GB			llama.cpp ›
665	Falcon-180B-Chat TII	40±2	local	~115GB		2k	Ollama ›
666	Code Llama 13B Instruct Meta	40±1	local	~15GB		16k	llama.cpp ›
667	MistralHermes-CodePro-7B-v1 beowolx	40±1	local	~6.1GB			llama.cpp ›
668	Phi-3-mini-4k-instruct (Apr 2024) Microsoft	40±2	local	~6GB			llama.cpp ›
669	CodeGemma 7B (v1.0) Google	40±1	local	~7.1GB			llama.cpp ›
671	Code Llama 13B Python Meta	40±1	local	~9.8GB		16k	llama.cpp ›
672	Apertus 8B Instruct (2509) Swiss AI	40±3	local	~6.3GB		66k	LM Studio ›
673	DeepSeek-V2 Lite Chat (16B-A2.4B) DeepSeek	40±1	local	~11GB		32k	LM Studio ›
674	Granite 3B Code Instruct (2K) IBM	40±1	local	~3.7GB			llama.cpp ›
675	Tulu 2 DPO 13B Allen Institute for AI	40±1	local	~9.8GB			llama.cpp ›
677	BioMistral-7B BioMistral	39±3	local	~6.1GB			llama.cpp ›
678	SOLAR 10.7B Instruct v1.0 Upstage	39±1	local	~8.3GB		4k	llama.cpp ›	+2
679	Code Tulu 2 7B Allen Institute for AI	39±2†	local	~16GB			how to run ›
680	Gemma 2 2B IT Google	39±1	local	~3.2GB		8k	LM Studio ›	+4
681	Yi-1.5 6B (base) 01.AI	39±3	local	~5.1GB		4k	LM Studio ›
682	Flan-T5 XXL Google	39±1	local	~49GB			how to run ›	+12
683	Dolphin 2.2.1 Mistral 7B Cognitive Computations	39±2	local	~5.9GB			llama.cpp ›
684	Phi-3.5-vision-instruct Microsoft	39±2	local	~10GB			how to run ›
685	Gemma 1.1 7B IT Google	39±1	local	~7.1GB		8k	llama.cpp ›	+2
686	Granite 3.1 2B Instruct IBM	39±1	local	~3.1GB		128k	llama.cpp ›
687	Granite 20B Code Base (8K) IBM	39±1	local	~15GB			llama.cpp ›
688	Qwen1.5 7B Chat Alibaba	39±1	local	~9.7GB		32k	llama.cpp ›	+4
689	Code Llama 7B Python Meta	39±1	local	~5.8GB		16k	llama.cpp ›
690	Granite 3.2 2B Instruct IBM	39±1	local	~3GB		131k	LM Studio ›
691	Falcon3 3B Instruct TII	39±3	local	~3.4GB		32k	LM Studio ›
692	Qwen2.5 Coder 0.5B Instruct Alibaba	39±1	local	~1GB		33k	LM Studio ›
693	Vicuna 13B v1.5 LMSYS	38±1	local	~9.8GB		4k	llama.cpp ›
694	Zephyr 7B alpha Hugging Face	38±1	local	~6.1GB			llama.cpp ›
695	Code Llama 13B (base) Meta	38±1	local	~9.2GB		16k	LM Studio ›	+1
696	Yi 1.5 6B Chat 01.AI	38±2	local	~4.4GB		4k	LM Studio ›
697	Qwen2-VL 2B Instruct Alibaba	38±1	local	~2.1GB		33k	LM Studio ›
698	WizardLM 13B V1.2 Microsoft	38±1	local	~9.8GB			llama.cpp ›
699	Qwen2.5 1.5B Instruct Alibaba	38±1	local	~1.8GB	~98 tok/s	33k	LM Studio ›	+6
700	Qwen3.5 0.8B Alibaba	38±1	local	~1.4GB	~94 tok/s	262k	llama.cpp ›	+2
701	Mistral 7B Instruct v0.2 Mistral	38±1	local	~5.8GB		32k	LM Studio ›	+9
702	Code Llama 7B Instruct Meta	38±1	local	~5.5GB		16k	LM Studio ›
703	Tulu 2 7B Allen Institute for AI	38±2	local	~5.8GB			llama.cpp ›
705	LFM2 700M Liquid AI	38±1	local	~1GB		33k	LM Studio ›
706	EXAONE 3.5 2.4B Instruct LG AI Research	38±2	local	~3GB		33k	LM Studio ›
707	xDAN-L1-Chat-RL-v1 xDAN-AI	38±1	local	~6.1GB			llama.cpp ›
710	Llama 3 8B (base) Meta	37±1	local	~6.6GB			llama.cpp ›
711	Mistral 7B v0.1 (base) Mistral AI	37±1	local	~5.8GB		8k	LM Studio ›	+4
712	Flan-T5 XL Google	37±1	local	~14GB			how to run ›	+8
713	Gemma 7B (base) Google	37±1	local	~7.1GB		8k	llama.cpp ›	+3
714	Tulu 2 DPO 7B Allen Institute for AI	37±1	local	~5.8GB			llama.cpp ›
715	Vicuna 13B v1.3 LMSYS	37±2	local	~29GB		2k	how to run ›
716	DeepSeek R1 Distill Qwen 1.5B DeepSeek	37±2	local	~1.9GB		131k	LM Studio ›	+1
717	Granite 3.1 3B-A800M Instruct IBM	37±2†	local	~3.6GB	~23 tok/s	131k	llama.cpp ›
718	Granite 3.0 3B-A800M Instruct IBM	37±1	local	~3.7GB		4k	llama.cpp ›
719	StarCoder2 7B BigCode	37±1	local	~5.4GB		16k	LM Studio ›
720	DeepSeek LLM 7B Chat DeepSeek	37±2†	local	~5.7GB			Ollama ›
722	StarCoder 15.5B BigCode	37±1	local	~11GB			Ollama ›
723	Eurus-2-7B-SFT PRIME-RL	37±3†	local	~6.4GB			llama.cpp ›
724	Falcon 40B (base) TII	36±1	local	~27GB		2k	Ollama ›	+2
725	Nous-Hermes-13b Nous Research	36±2	local	~9.8GB		2k	llama.cpp ›
727	Code Llama 7B (base) Meta	36±1	local	~5.5GB		16k	LM Studio ›	+2
728	Llama 2 13B Chat Meta	36±1	local	~15GB		4k	llama.cpp ›	+10
729	Llama 2 13B (base) Meta	36±1	local	~9.3GB		4k	Ollama ›	+4
730	Zephyr 7B beta Hugging Face	36±1	local	~6.1GB			llama.cpp ›	+3
731	Guanaco 33B University of Washington	35±1	local	~70GB			how to run ›
732	Qwen3 0.6B Alibaba	35±1	local	~1.9GB	~182 tok/s	33k	LM Studio ›	+4
733	StarCoder2 3B BigCode	35±1	local	~3.4GB		16k	llama.cpp ›
734	Vicuna 7B v1.5 LMSYS	35±1	local	~5.8GB		4k	llama.cpp ›	+1
735	Granite 3B Code Base (2K) IBM	35±1	local	~3.7GB			llama.cpp ›
736	StarCoderBase 15.5B BigCode	35±1	local	~12GB			llama.cpp ›
738	OLMo 2 1B Instruct (0425) Allen Institute for AI	35±1	local	~2.4GB		4k	llama.cpp ›
739	ChatGLM3 6B 128K Zhipu AI (THUDM)	35±4	local	~5.8GB		131k	llama.cpp ›
740	Orca 2 13B Microsoft	35±3	local	~9.8GB		4k	llama.cpp ›
741	Vicuna 7B v1.3 LMSYS	35±2	local	~5.8GB		2k	llama.cpp ›
742	Stable Code 3B Stability AI	35±1	local	~4.8GB		16k	llama.cpp ›
744	Qwen1.5 4B Chat Alibaba	34±2	local	~4.1GB		32k	llama.cpp ›
745	DeepSeek-Coder 1.3B Base DeepSeek	34±1	local	~2.4GB		16k	Ollama ›
746	Gemma 7B IT Google	34±1	local	~7.1GB		8k	llama.cpp ›	+5
747	Mistral 7B Instruct v0.1 Mistral	34±1	local	~5.9GB		8k	llama.cpp ›	+3
750	CodeGemma 2B (v1.0) Google	34±1	local	~3.2GB			llama.cpp ›
751	Qwen2 1.5B Instruct Alibaba	34±1	local	~1.8GB		32k	LM Studio ›	+1
752	Ghost 7B Alpha Ghost X	34±2	local	~6.1GB			llama.cpp ›
753	Falcon-H1 1.5B Instruct TII	33±4	local	~2.4GB		131k	LM Studio ›
754	Gemma 3 1B Google	33±1	local	~1.7GB	~98 tok/s	32k	LM Studio ›	+7
755	SmolLM2 1.7B Instruct Hugging Face	33±1	local	~3GB		8k	LM Studio ›
756	Qwen-7B (base) Alibaba	33±1	local	~6.6GB		33k	llama.cpp ›
757	LFM2 350M Liquid AI	33±1	local	~1.7GB		33k	LM Studio ›
761	StarCoderBase 7B BigCode	33±1	local	~6GB			Ollama ›
762	Llama 3.2 1B Instruct Meta	33±1	api $0.020/M +local	~1.6GB	~28 tok/s	131k	novita.ai ↗	+13
763	Phi-1.5 (1.3B) Microsoft	33±1	local	~2.4GB		2k	llama.cpp ›
765	Llama 2 7B Chat Meta	32±1	local	~8.8GB		4k	llama.cpp ›	+13
766	Qwen-7B-Chat Alibaba	32±1	local	~6.6GB		33k	llama.cpp ›
767	OLMo 7B Instruct Ai2	32±2	local	~16GB			how to run ›
768	Llama 2 7B (base) Meta	32±1	local	~8.5GB		4k	Ollama ›	+6
771	Flan-T5 Large Google	32±1	local	~4.8GB			how to run ›	+5
772	Granite 3.1 1B-A400M Instruct IBM	31±2†	local	~2.3GB	~20 tok/s	131k	llama.cpp ›
773	Falcon 40B Instruct TII	31±2	local	~27GB			Ollama ›
774	Gemma 2B (base) Google	31±1	local	~3.2GB		8k	llama.cpp ›	+3
775	Gemma 1.1 2B IT Google	31±1	local	~3.2GB		8k	llama.cpp ›	+1
776	Falcon 7B (base) TII	31±1	local	~5.9GB		2k	Ollama ›	+4
778	Baichuan2-7B-Base Baichuan	30±1	local	~17GB			how to run ›
779	NexusRaven V2 13B Nexusflow	30±3	local	~9.3GB		16k	Ollama ›
782	ChatGLM3-6B Zhipu AI (THUDM)	30±2	local	~5.8GB			llama.cpp ›
783	Qwen2.5 0.5B Instruct Alibaba	30±2	local	~1GB		33k	LM Studio ›	+1
784	OLMo 7B Allen Institute for AI	30±1†	local	~30GB		2k	how to run ›
785	OpenLLaMA 7B v2 OpenLM Research	30±1	local	~9.1GB			llama.cpp ›
786	StableLM-3B-4E1T Stability AI	30±1	local	~3.3GB		4k	llama.cpp ›
787	StarCoderBase 3B BigCode	30±1	local	~3.4GB			Ollama ›
788	Granite 3.0 1B-A400M Instruct IBM	30±1	local	~2.3GB		4k	llama.cpp ›
789	OPT-66B Meta	30±1	local	~141GB		2k	how to run ›	+2
791	StableLM-Base-Alpha 7B v2 Stability AI	30±1	local	~30GB		4k	how to run ›
792	GPT4All-13B-Snoozy Nomic AI	29±2	local	~56GB			how to run ›
794	EuroLLM 1.7B Instruct UTTER (EuroLLM)	29±3	local	~2.4GB		4k	LM Studio ›
795	Granite 4.0 H 350M IBM	29±1†	local	~1.7GB		33k	LM Studio ›
797	BLOOM 176B BigScience	29±1	local	~372GB		2k	how to run ›	+1
798	Gemma 2B IT Google	29±1	local	~3.1GB		8k	llama.cpp ›	+3
799	Granite 4.0 350M IBM	28±1	local	~1.7GB		32k	LM Studio ›
800	OPT-30B Meta	28±1	local	~64GB		2k	how to run ›	+5
802	GPT-NeoX-20B EleutherAI	28±1	local	~14GB		2k	LM Studio ›	+1
803	Qwen2 0.5B Instruct Alibaba	28±2	local	~1GB		32k	LM Studio ›
804	RedPajama-INCITE 7B Base Together	28±1	local	~16GB		2k	how to run ›
805	ChatGLM2-6B Zhipu AI (THUDM)	28±2	local	~15GB			how to run ›
809	Qwen1.5 1.8B Chat Alibaba	26±2	local	~2.7GB		33k	Ollama ›
811	Falcon3 1B Instruct TII	26±4	local	~2.4GB		8k	LM Studio ›
812	OPT-13B (13 billion parameters) Meta	26±1	local	~29GB		2k	how to run ›	+4
814	Falcon 7B Instruct TII	26±2	local	~5.9GB			Ollama ›	+1
815	OpenAssistant SFT-4 Pythia 12B LAION (OpenAssistant)	25±1	local	~9.5GB			llama.cpp ›	+1
817	RedPajama-INCITE Base 3B v1 Together	25±3	local	~3.4GB		2k	llama.cpp ›
818	Pythia 12B EleutherAI	25±1	local	~9.5GB		2k	llama.cpp ›	+1
819	Flan-T5 Base Google	25±1	local	~2.5GB			how to run ›	+3
820	OpenLLaMA 3B v2 OpenLM Research	25±1	local	~4.2GB			llama.cpp ›
821	GPT-J 6B EleutherAI	25±1	local	~14GB		2k	how to run ›	+6
822	ChatGLM-6B Zhipu AI (THUDM)	25±1	local	~16GB			how to run ›	+3
823	StableLM-Base-Alpha 3B v2 Stability AI	24±1	local	~13GB		4k	how to run ›
824	OPT-6.7B Meta	24±1	local	~16GB		2k	how to run ›	+5
827	Qwen1.5 0.5B Chat Alibaba	23±3	local	~1.7GB		33k	LM Studio ›
828	Pythia 6.9B EleutherAI	23±1	local	~6.1GB		2k	llama.cpp ›	+2
830	Phi-1 (1.3B) Microsoft	22±4	local	~2.4GB		2k	llama.cpp ›
832	SmolLM2 360M Instruct Hugging Face	22±3	local	~1GB		8k	LM Studio ›
833	FastChat-T5 3B LMSYS	21±2	local	~8.5GB			how to run ›
834	BLOOM 7.1B BigScience	21±1	local	~7.1GB		2k	llama.cpp ›	+7
835	OLMo 1B Allen Institute for AI	21±2†	local	~2.2GB		2k	llama.cpp ›
836	Pythia 2.8B EleutherAI	20±1	local	~3.4GB		2k	llama.cpp ›
837	Pythia 2.8B (deduped) EleutherAI	20±1	local	~3.4GB		2k	llama.cpp ›
838	Falcon-RW 1B TII	20±2	local	~4.3GB		2k	how to run ›
839	OPT-2.7B Meta	19±1	local	~7.1GB		2k	how to run ›	+3
840	GPT-Neo 2.7B EleutherAI	18±1	local	~13GB		2k	how to run ›	+8
841	StableLM-Tuned-Alpha 7B Stability AI	18±2	local	~6.9GB			llama.cpp ›
843	T5 Large Google	17±2	local	~4.6GB		1k	how to run ›
844	Gemma 3 270M IT Google	16±4	local	~1.1GB		32k	LM Studio ›
845	BLOOM 3B BigScience	16±1	local	~4GB		2k	llama.cpp ›	+2
846	Flan-T5 Small Google	16±3	local	~1.8GB			how to run ›
849	OPT-1.3B Meta	14±1	local	~4.3GB		2k	how to run ›	+5
850	T5 Base Google	14±3	local	~2.4GB		1k	how to run ›
851	Pythia 1.4B EleutherAI	14±1	local	~2.4GB		2k	llama.cpp ›
853	Pythia 1B EleutherAI	12±1	local	~2.2GB		2k	llama.cpp ›
855	GPT-Neo 1.3B EleutherAI	12±2	local	~7.1GB		2k	how to run ›	+3
856	T5 Small Google	12±3	local	~1.8GB		1k	how to run ›
857	BLOOM 1.7B BigScience	11±1	local	~3.1GB		2k	llama.cpp ›	+1
858	Zyte-1B venkycs (aihub-app)	10±3	local	~2.2GB			llama.cpp ›
860	StableLM-Base-Alpha 7B Stability AI	8±1	local	~35GB		4k	how to run ›
861	Pythia 410M EleutherAI	8±2	local	~1.8GB		2k	llama.cpp ›
863	BLOOM 1.1B BigScience	7±1	local	~2.6GB		2k	llama.cpp ›
864	GPT-2 XL (1.5B) OpenAI	7±1	local	~2.8GB	~75 tok/s	1k	llama.cpp ›	+9
867	OPT-350M Meta	1±1	local	~2.2GB		2k	how to run ›	+1
869	StableLM-Base-Alpha 3B Stability AI	1±2	local	~17GB		4k	how to run ›
870	BLOOM 560M BigScience	0±1	local	~2.1GB		2k	llama.cpp ›	+1
871	Pythia 160M EleutherAI	-2±2	local	~1.6GB		2k	llama.cpp ›
872	GPT-2 Large (774M) OpenAI	-3±1	local	~2GB	~137 tok/s	1k	llama.cpp ›	+3
873	OPT-125M Meta	-3±1	local	~1.8GB		2k	how to run ›	+1
874	GPT-Neo 125M EleutherAI	-3±2	local	~2.1GB		2k	how to run ›
875	GPT-2 Medium (355M) OpenAI	-5±1	local	~1.8GB		1k	llama.cpp ›	+3
876	Pythia 70M EleutherAI	-13±2	local	~1.6GB		2k	llama.cpp ›
877	GPT-2 Small (124M) OpenAI	-13±2	local	~1.6GB		1k	LM Studio ›	+3
—	K2 Horizon 7B Institute of Foundation Models (MBZUAI)	not scored yet	local	~8GB		524k	llama.cpp ›

+ 279 more don't fit your machine or budget · show them

+ 220 models with fewer than 3 results left off, since one or two tests say little about overall ability · show them

907 of 1212 catalogue models listed here · sources and licences

* new or thinly tested: ranked on too few benchmark results · † only the maker’s own numbers so far · ~RAM is estimated; full context may need more memory