Supported models

A curated set of open-weight foundation models is available, optimized to run on the MAX framework. For the live list of available models and their identifiers, see the Models page in the console. Models are available through two kinds of endpoints:

Shared endpoints — serverless, pay-as-you-go inference on a curated set of models. No deployment or setup required; call the endpoint directly with your API key.
Dedicated endpoints — a dedicated deployment of a model from Modular Cloud’s broader catalog, provisioned exclusively for your organization. See Create a deployment.

Check the tables below to see which models support function calling and reasoning, and which API endpoint (/v1/chat/completions or /v1/responses) each model uses.

Shared endpoints

These models run on shared, multi-tenant infrastructure and are available out of the box to every organization.

Model	Provider	Modalities	Endpoint	Context	Parameters	Function calling	Reasoning
Gemma 4 26B-A4B `google/gemma-4-26B-A4B-it`	Google DeepMind	Text, Vision	`/v1/chat/completions`	128K	4B active / 26B total	✓	—
Gemma 4 31B `google/gemma-4-31B-it`	Google DeepMind	Text, Vision	`/v1/chat/completions`	128K	31B	✓	—
Kimi K2.5 `moonshotai/Kimi-K2.5`	Moonshot AI	Text, Vision	`/v1/chat/completions`	256K	32B active / ~1T total	✓	✓
FLUX.2-dev `black-forest-labs/FLUX.2-dev`	Black Forest Labs	Image	`/v1/responses`	—	12B	—	—
FLUX.2-klein-4B `black-forest-labs/FLUX.2-klein-4B`	Black Forest Labs	Image	`/v1/responses`	—	4B	—	—
FLUX.2-klein-9B `black-forest-labs/FLUX.2-klein-9B`	Black Forest Labs	Image	`/v1/responses`	—	9B	—	—

Dedicated endpoints

Custom dedicated deployments are available for these models. Contact us for more info on dedicated deployments.

Model	Provider	Modalities	Endpoint	Context	Parameters	Function calling	Reasoning
DeepSeek R1 `deepseek-ai/DeepSeek-R1`	DeepSeek	Text	`/v1/chat/completions`	128K	37B active / 671B total	—	✓
DeepSeek R1-0528 `deepseek-ai/DeepSeek-R1-0528`	DeepSeek	Text	`/v1/chat/completions`	128K	37B active / 671B total	—	✓
DeepSeek R1 Distill Llama 70B `deepseek-ai/DeepSeek-R1-Distill-Llama-70B`	DeepSeek	Text	`/v1/chat/completions`	128K	70B	—	✓
DeepSeek V3.1 `deepseek-ai/DeepSeek-V3.1`	DeepSeek	Text	`/v1/chat/completions`	128K	37B active / 671B total	✓	✓
DeepSeek V3.2 `deepseek-ai/DeepSeek-V3.2`	DeepSeek	Text	`/v1/chat/completions`	128K	37B active / 685B total	✓	—
Gemma 3 12B `google/gemma-3-12b-it`	Google DeepMind	Text, Vision	`/v1/chat/completions`	128K	12B	—	—
Gemma 3 27B `google/gemma-3-27b-it`	Google DeepMind	Text, Vision	`/v1/chat/completions`	128K	27B	—	—
EXAONE 4.0 32B `LGAI-EXAONE/EXAONE-4.0-32B`	LG AI Research	Text	`/v1/chat/completions`	128K	32B	✓	✓
Llama 3.3 70B `meta-llama/Llama-3.3-70B-Instruct`	Meta	Text	`/v1/chat/completions`	128K	70B	✓	—
Llama 4 Maverick `meta-llama/Llama-4-Maverick-17B-128E-Instruct`	Meta	Text, Vision	`/v1/chat/completions`	1M	17B active / 400B total	✓	—
Llama 4 Scout `meta-llama/Llama-4-Scout-17B-16E-Instruct`	Meta	Text, Vision	`/v1/chat/completions`	10M	17B active / 109B total	✓	—
Llama 3.1 405B `meta-llama/Meta-Llama-3.1-405B-Instruct`	Meta	Text	`/v1/chat/completions`	128K	405B	✓	—
Phi-4 `microsoft/phi-4`	Microsoft	Text	`/v1/chat/completions`	16K	14B	—	—
MiniMax M2.5 `MiniMaxAI/MiniMax-M2.5`	MiniMax	Text	`/v1/chat/completions`	200K	10B active / 230B total	—	—
Ministral 14B `mistralai/Ministral-3-14B-Instruct-2512`	Mistral AI	Text, Vision	`/v1/chat/completions`	256K	14B	✓	✓
Mistral Large 3 `mistralai/Mistral-Large-3-675B-Instruct-2512`	Mistral AI	Text, Vision	`/v1/chat/completions`	256K	41B active / 675B total	✓	—
Mistral Small 3.1 24B `mistralai/Mistral-Small-3.1-24B-Instruct-2503`	Mistral AI	Text, Vision	`/v1/chat/completions`	128K	24B	✓	—
Nemotron 3 Nano `nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-FP8`	NVIDIA	Text	`/v1/chat/completions`	1M	3.2B active / 31.6B total	—	✓
gpt-oss-120b `openai/gpt-oss-120b`	OpenAI	Text	`/v1/chat/completions`	128K	5.1B active / 117B total	✓	✓
gpt-oss-20b `openai/gpt-oss-20b`	OpenAI	Text	`/v1/chat/completions`	128K	3.6B active / 21B total	✓	✓
Qwen2.5 72B `Qwen/Qwen2.5-72B-Instruct`	Alibaba	Text	`/v1/chat/completions`	128K	72B	✓	—
Qwen3-235B-A22B `Qwen/Qwen3-235B-A22B-Instruct-2507`	Alibaba	Text	`/v1/chat/completions`	262K	22B active / 235B total	✓	✓
Qwen3 30B-A3B `Qwen/Qwen3-30B-A3B`	Alibaba	Text	`/v1/chat/completions`	262K	3.3B active / 30.5B total	✓	✓
Qwen3-Coder-480B-A35B `Qwen/Qwen3-Coder-480B-A35B-Instruct`	Alibaba	Text	`/v1/chat/completions`	262K	35B active / 480B total	✓	✓
Qwen3-Omni-30B-A3B `Qwen/Qwen3-Omni-30B-A3B-Instruct`	Alibaba	Text, Audio, Vision	`/v1/chat/completions`	262K	3B active / 30B total	—	✓
Qwen3-VL-30B-A3B `Qwen/Qwen3-VL-30B-A3B-Instruct`	Alibaba	Text, Vision	`/v1/chat/completions`	256K	3B active / 30B total	—	✓
Qwen3-VL-4B `Qwen/Qwen3-VL-4B-Instruct`	Alibaba	Text, Vision	`/v1/chat/completions`	256K	4B	—	✓
Qwen3-VL-8B `Qwen/Qwen3-VL-8B-Instruct`	Alibaba	Text, Vision	`/v1/chat/completions`	256K	8B	—	✓
Qwen3.5-397B-A17B `Qwen/Qwen3.5-397B-A17B`	Alibaba	Text, Vision	`/v1/chat/completions`	262K	17B active / 397B total	✓	✓
MiMo-V2-Flash `XiaomiMiMo/MiMo-V2-Flash`	Xiaomi	Text	`/v1/chat/completions`	256K	15B active / 309B total	—	✓
Wan2.2-T2V-A14B `Wan-AI/Wan2.2-T2V-A14B-Diffusers`	Wan-AI	Video	`/v1/responses`	—	14B	—	—
GLM-4.7 `zai-org/GLM-4.7`	Zhipu AI	Text, Vision, Audio	`/v1/chat/completions`	200K	32B active / 355B total	✓	✓
GLM-5 `zai-org/GLM-5`	Zhipu AI	Text	`/v1/chat/completions`	200K	44B active / 744B total	✓	✓

Model availability changes frequently as Modular adds and promotes new models. Always refer to the console for the current list.

Using model IDs

Pass the model identifier shown in the tables above as the model parameter in API requests. For example:

"model": "meta-llama/Llama-4-Maverick-17B-128E-Instruct"

Inference

Deployments

Administration

Shared endpoints

Dedicated endpoints

Using model IDs

​Shared endpoints

​Dedicated endpoints

​Using model IDs

Shared endpoints

Dedicated endpoints

Using model IDs