Catalogue snapshot 2026-10-03 / last 30 days

New models, and what runs them

A model is released and the first question is what it takes to run. Each one added here is sized against every catalogued card the day it lands.

Follow new models as they are added: the Atom feed.

Model	Smallest card that holds it (Q4_K_M)	Cheapest rented	Per token
AliceAI-Foundation-80B-A3B-Baseyandex · added 2026-10-03	Apple M1 Max (64 GB) · ~86 tok/s	4× RTX 5060 Ti 16GB · $0.51/h	—
MiMo-V2.6-Pro-MOPDXiaomiMiMo · added 2026-10-03	No single card a person can buy	8× RTX PRO 6000 · $16.54/h	—
MiMo-V2.6-Flash-MOPDXiaomiMiMo · added 2026-10-03	Apple M5 Ultra (512 GB) · ~70 tok/s	8× L40S · $3.73/h	—
MiMo-V2.6-Distill-Qwen-9BXiaomiMiMo · added 2026-10-03	NVIDIA GeForce RTX 3070 Ti (8 GB) · ~85 tok/s	RTX 3060 · $0.036/h	—
Swift-1.5-Qwen3.8-27bukisai · added 2026-10-03	AMD Radeon RX 7900 XT (20 GB) · ~42 tok/s	2× RTX 3060 · $0.11/h	—
llm-jp-4.1-8b-thinkingllm-jp · added 2026-10-03	NVIDIA GeForce RTX 3070 Ti (8 GB) · ~70 tok/s	RTX 3060 · $0.036/h	—
llm-jp-4.1-33b-thinkingllm-jp · added 2026-10-03	AMD Radeon™ RX 7900 XTX (24 GB) · ~34 tok/s	2× RTX 3060 · $0.11/h	—
llm-jp-4.1-32b-a3b-thinkingllm-jp · added 2026-10-03	AMD Radeon™ RX 7900 XTX (24 GB) · ~281 tok/s	2× RTX 3060 · $0.11/h	—
Holo4-35B-A3BHcompany · added 2026-10-03	AMD Radeon™ RX 7900 XTX (24 GB) · ~403 tok/s	2× RTX 3060 · $0.11/h	—
Holo4-27BHcompany · added 2026-10-03	AMD Radeon RX 7900 XT (20 GB) · ~42 tok/s	2× RTX 3060 · $0.11/h	—
Qwen3.8-35B-A3B-Distillempero-ai · added 2026-10-03	AMD Radeon™ RX 7900 XTX (24 GB) · ~403 tok/s	2× RTX 3060 · $0.11/h	—
LensVLM-9Bapple · added 2026-10-03	NVIDIA GeForce RTX 3070 Ti (8 GB) · ~85 tok/s	RTX 3060 · $0.036/h	—
Kolibri-1-BF16Aleph-Alpha · added 2026-10-03	Apple M1 Max (64 GB) · ~81 tok/s	4× RTX 5060 Ti 16GB · $0.51/h	—
Edge0-35B-A3B-previewEdge0 · added 2026-10-03	AMD Radeon™ RX 7900 XTX (24 GB) · ~403 tok/s	2× RTX 3060 · $0.11/h	—
Mistral-Large-3-675B-Instruct-2512Mistral AI · added 2026-10-03	Apple M5 Ultra (512 GB) · ~41 tok/s	2× B300 · $15.78/h	—
MiMo-V2.6-Flash-RLXiaomiMiMo · added 2026-10-03	Apple M5 Ultra (512 GB) · ~70 tok/s	8× L40S · $3.73/h	—
MiMo-V2.6-Pro-RLXiaomiMiMo · added 2026-10-03	No single card a person can buy	8× RTX PRO 6000 · $16.54/h	—
JiRackUltra_1bCMSManhattan · added 2026-09-24	NVIDIA GeForce RTX 3050 Ti Laptop GPU (4 GB) · ~107 tok/s	RTX 3060 · $0.036/h	—
K2-Horizon-375B-A23BIFM · added 2026-09-24	Apple M5 Ultra (512 GB) · ~49 tok/s	8× L40S · $3.73/h	—
K2-Horizon-3.7BIFM · added 2026-09-24	NVIDIA GeForce RTX 2060 (6 GB) · ~57 tok/s	RTX 3060 · $0.036/h	—
Hemmingway-1Altworld · added 2026-09-24	AMD Radeon RX 7900 XT (20 GB) · ~42 tok/s	2× RTX 3060 · $0.11/h	—
GLM-4.6V-Flashzai-org · added 2026-09-20	NVIDIA GeForce RTX 3070 Ti (8 GB) · ~72 tok/s	RTX 3060 · $0.036/h	—
Xing4.0-29B-A4BXingChen-AGI · added 2026-09-20	AMD Radeon™ RX 7900 XTX (24 GB) · ~288 tok/s	2× RTX 3060 · $0.11/h	—
MiMo-V2-FlashXiaomiMiMo · added 2026-09-20	Apple M5 Ultra (512 GB) · ~70 tok/s	8× L40S · $3.73/h	—
Spark-X2.5-1.7BXHToken · added 2026-09-20	NVIDIA GeForce RTX 3050 Ti Laptop GPU (4 GB) · ~107 tok/s	RTX 3060 · $0.036/h	—
Meta-Llama-3.1-70B-InstructMeta · added 2026-09-20	NVIDIA RTX PRO 5000 Blackwell 48GB (48 GB) · ~20 tok/s	2× RTX 3090 · $0.24/h	—
Llama-4-Scout-17B-16E-InstructMeta · added 2026-09-20	NVIDIA RTX PRO 5000 Blackwell 72GB (72 GB) · ~116 tok/s	8× RTX 3060 · $0.59/h	$0.30/M
Llama-4-Maverick-17B-128E-InstructMeta · added 2026-09-20	Apple M5 Ultra (512 GB) · ~78 tok/s	8× L40S · $3.73/h	$0.65/M
Llama-3.3-70B-InstructMeta · added 2026-09-20	NVIDIA RTX PRO 5000 Blackwell 48GB (48 GB) · ~20 tok/s	2× RTX 3090 · $0.24/h	$0.32/M
Llama-3.2-3B-InstructMeta · added 2026-09-20	NVIDIA GeForce RTX 3050 Ti Laptop GPU (4 GB) · ~44 tok/s	RTX 3060 · $0.036/h	$0.33/M
Llama-3.2-1B-InstructMeta · added 2026-09-20	NVIDIA GeForce RTX 3050 Ti Laptop GPU (4 GB) · ~122 tok/s	RTX 3060 · $0.036/h	$0.02/M
gemma-3-4b-itGoogle · added 2026-09-20	NVIDIA GeForce RTX 3050 Ti Laptop GPU (4 GB) · ~50 tok/s	RTX 3060 · $0.036/h	$0.10/M
gemma-3-27b-itGoogle · added 2026-09-20	AMD Radeon RX 7900 XT (20 GB) · ~37 tok/s	2× RTX 3060 · $0.11/h	$0.20/M
gemma-3-1b-itGoogle · added 2026-09-20	NVIDIA GeForce RTX 3050 Ti Laptop GPU (4 GB) · ~191 tok/s	RTX 3060 · $0.036/h	—
gemma-3-12b-itGoogle · added 2026-09-20	NVIDIA GeForce RTX 3080 (10 GB) · ~67 tok/s	RTX 3060 · $0.036/h	$0.10/M
gemma-2-9b-itGoogle · added 2026-09-20	NVIDIA GeForce RTX 3080 (10 GB) · ~72 tok/s	RTX 3060 · $0.036/h	—
gemma-2-2b-itGoogle · added 2026-09-20	NVIDIA GeForce RTX 3050 Ti Laptop GPU (4 GB) · ~61 tok/s	RTX 3060 · $0.036/h	—
gemma-2-27b-itGoogle · added 2026-09-20	AMD Radeon RX 7900 XT (20 GB) · ~34 tok/s	2× RTX 3060 · $0.11/h	$0.65/M
NeoHorse-1-9BTokenRhythm · added 2026-09-20	NVIDIA GeForce RTX 3070 Ti (8 GB) · ~85 tok/s	RTX 3060 · $0.036/h	—
Apertus-70B-Instruct-2509swiss-ai · added 2026-09-20	NVIDIA RTX PRO 5000 Blackwell 48GB (48 GB) · ~15 tok/s	2× RTX 3090 · $0.24/h	—
Supra2-100M-InstructSupraLabs · added 2026-09-20	Not sized at Q4_K_M	RTX 3060 · $0.036/h	—
Step-3.7-Flashstepfun-ai · added 2026-09-20	AMD Ryzen AI Max+ PRO 495 with Radeon 8065S (160 GB) · ~18 tok/s	4× L40S · $1.87/h	$1.15/M
Step-3.5-Flashstepfun-ai · added 2026-09-20	Apple M1 Ultra (128 GB) · ~47 tok/s	4× L40S · $1.87/h	$0.30/M
sarvam-105bsarvamai · added 2026-09-20	NVIDIA RTX PRO 5000 Blackwell 72GB (72 GB) · ~133 tok/s	8× RTX 3060 · $0.59/h	—
Qwen3-VL-4B-InstructQwen · added 2026-09-20	NVIDIA GeForce RTX 2060 (6 GB) · ~62 tok/s	RTX 3060 · $0.036/h	—
Qwen3-VL-2B-InstructQwen · added 2026-09-20	NVIDIA GeForce RTX 3050 Ti Laptop GPU (4 GB) · ~64 tok/s	RTX 3060 · $0.036/h	—
Qwen3.5-2BQwen · added 2026-09-20	NVIDIA GeForce RTX 3050 Ti Laptop GPU (4 GB) · ~106 tok/s	RTX 3060 · $0.036/h	—
Qwen3.5-0.8BQwen · added 2026-09-20	NVIDIA GeForce RTX 3050 Ti Laptop GPU (4 GB) · ~232 tok/s	RTX 3060 · $0.036/h	—
Qwen3-4B-Instruct-2507Qwen · added 2026-09-20	NVIDIA GeForce RTX 2060 (6 GB) · ~62 tok/s	RTX 3060 · $0.036/h	—
Qwen3-4B-BaseQwen · added 2026-09-20	NVIDIA GeForce RTX 2060 (6 GB) · ~62 tok/s	RTX 3060 · $0.036/h	—
Qwen3-1.7B-BaseQwen · added 2026-09-20	NVIDIA GeForce RTX 3050 Ti Laptop GPU (4 GB) · ~64 tok/s	RTX 3060 · $0.036/h	—
Qwen2.5-VL-7B-InstructQwen · added 2026-09-20	NVIDIA GeForce RTX 3070 Ti (8 GB) · ~85 tok/s	RTX 3060 · $0.036/h	—
Qwen2.5-VL-3B-InstructQwen · added 2026-09-20	NVIDIA GeForce RTX 3050 Ti Laptop GPU (4 GB) · ~59 tok/s	RTX 3060 · $0.036/h	—
Qwen2.5-VL-32B-InstructQwen · added 2026-09-20	AMD Radeon™ RX 7900 XTX (24 GB) · ~35 tok/s	RTX 3090 · $0.12/h	—
Qwen2.5-Coder-7B-InstructQwen · added 2026-09-20	NVIDIA GeForce RTX 2060 (6 GB) · ~47 tok/s	RTX 3060 · $0.036/h	—
Qwen2.5-Coder-3B-InstructQwen · added 2026-09-20	NVIDIA GeForce RTX 3050 Ti Laptop GPU (4 GB) · ~59 tok/s	RTX 3060 · $0.036/h	—
Qwen2.5-Coder-14B-InstructQwen · added 2026-09-20	NVIDIA GeForce RTX 3080 Ti (12 GB) · ~61 tok/s	RTX 3060 · $0.036/h	—
Qwen2.5-Coder-1.5B-InstructQwen · added 2026-09-20	NVIDIA GeForce RTX 3050 Ti Laptop GPU (4 GB) · ~107 tok/s	RTX 3060 · $0.036/h	—
Qwen2.5-3B-InstructQwen · added 2026-09-20	NVIDIA GeForce RTX 3050 Ti Laptop GPU (4 GB) · ~59 tok/s	RTX 3060 · $0.036/h	—
Qwen2.5-14B-InstructQwen · added 2026-09-20	NVIDIA GeForce RTX 3080 Ti (12 GB) · ~61 tok/s	RTX 3060 · $0.036/h	—
Qwen2.5-1.5B-InstructQwen · added 2026-09-20	NVIDIA GeForce RTX 3050 Ti Laptop GPU (4 GB) · ~107 tok/s	RTX 3060 · $0.036/h	—
Qwen2.5-0.5B-InstructQwen · added 2026-09-20	NVIDIA GeForce RTX 3050 Ti Laptop GPU (4 GB) · ~305 tok/s	RTX 3060 · $0.036/h	—
Qwen2.5-0.5BQwen · added 2026-09-20	NVIDIA GeForce RTX 3050 Ti Laptop GPU (4 GB) · ~305 tok/s	RTX 3060 · $0.036/h	—
NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16NVIDIA · added 2026-09-20	Apple M5 Ultra (512 GB) · ~3 tok/s	8× L40S · $3.73/h	$2.40/M
Kimi-Linear-48B-A3B-Instructmoonshotai · added 2026-09-20	NVIDIA GeForce RTX 5090 (32 GB) · ~652 tok/s	4× RTX 3060 · $0.21/h	—
Mixtral-8x7B-Instruct-v0.1Mistral AI · added 2026-09-20	NVIDIA GeForce RTX 5090 (32 GB) · ~139 tok/s	4× RTX 3060 · $0.21/h	—
Mistral-7B-Instruct-v0.3Mistral AI · added 2026-09-20	NVIDIA GeForce RTX 3070 Ti (8 GB) · ~77 tok/s	RTX 3060 · $0.036/h	—
EXAONE-4.0-32BLGAI-EXAONE · added 2026-09-20	AMD Radeon™ RX 7900 XTX (24 GB) · ~37 tok/s	2× RTX 3060 · $0.11/h	—
EXAONE-4.0-1.2BLGAI-EXAONE · added 2026-09-20	NVIDIA GeForce RTX 3050 Ti Laptop GPU (4 GB) · ~99 tok/s	RTX 3060 · $0.036/h	—
Atria-Dawn-Previewinternlm · added 2026-09-20	Apple M5 Ultra (512 GB) · ~37 tok/s	2× B300 · $15.78/h	—
K2-Horizon-7BIFM · added 2026-09-20	NVIDIA GeForce RTX 3070 Ti (8 GB) · ~66 tok/s	RTX 3060 · $0.036/h	—
K2-Horizon-0.9BIFM · added 2026-09-20	NVIDIA GeForce RTX 3050 Ti Laptop GPU (4 GB) · ~121 tok/s	RTX 3060 · $0.036/h	—
granite-4.1-3bIBM · added 2026-09-20	NVIDIA GeForce RTX 3050 Ti Laptop GPU (4 GB) · ~47 tok/s	RTX 3060 · $0.036/h	—
SmolLM2-1.7B-InstructHuggingFaceTB · added 2026-09-20	NVIDIA GeForce RTX 3050 Ti Laptop GPU (4 GB) · ~49 tok/s	RTX 3060 · $0.036/h	—
gemma-4-E4B-itGoogle · added 2026-09-20	NVIDIA GeForce RTX 2060 (6 GB) · ~76 tok/s	RTX 3060 · $0.036/h	—
gemma-4-E2B-itGoogle · added 2026-09-20	NVIDIA GeForce RTX 2060 (6 GB) · ~200 tok/s	RTX 3060 · $0.036/h	—
North-Micro-Vision-InstructCohereLabs · added 2026-09-20	NVIDIA GeForce RTX 3050 Ti Laptop GPU (4 GB) · ~70 tok/s	RTX 3060 · $0.036/h	—
OLMo-2-0325-32B-Instructallenai · added 2026-09-20	Not sized at Q4_K_M	2× RTX 3060 · $0.11/h	—
GigaChat3.5-432B-A28B-Reasoningai-sage · added 2026-09-20	Apple M5 Ultra (512 GB) · ~60 tok/s	8× L40S · $3.73/h	—

Speeds are single-stream estimates; rental prices are from the providers’ own APIs at the snapshot time and are live on each model page. Some rental links on those pages are referral links; how we rank.