AI models on Halo

Access these models with no subscription — pay per prompt in USDC, no account, no phone number. Verifiable and private.

165 models live on the network

Every model here is served right now by independent operators — pick any one, no subscription, pay per prompt.

OpenAI16

gpt-4.1-minigpt-4.1-nanogpt-4o-minigpt-5.2-codexgpt-5.3-chatgpt-5.3-codexgpt-5.4gpt-5.4-minigpt-5.4-nanogpt-5.4-progpt-5.5gpt-5.5-progpt-chat-latestgpt-latestgpt-mini-latestgpt-oss-120b

Anthropic13

claude-3-haikuclaude-fable-5claude-fable-latestclaude-haiku-latestclaude-opus-4.6claude-opus-4.7claude-opus-4.7-fastclaude-opus-4.8claude-opus-4.8-fastclaude-opus-latestclaude-sonnet-4.6claude-sonnet-5claude-sonnet-latest

Google16

gemini-3-flash-previewgemini-3-pro-imagegemini-3.1-flash-imagegemini-3.1-flash-litegemini-3.1-flash-lite-imagegemini-3.1-flash-lite-previewgemini-3.1-pro-previewgemini-3.1-pro-preview-customtoolsgemini-3.5-flashgemini-flash-latestgemini-pro-latestgemma-2-27b-itgemma-2-9b-itgemma-4-26b-a4b-itgemma-4-31B-itgemma-4-31b-it

DeepSeek4

deepseek-chatdeepseek-v3.2deepseek-v4-flashdeepseek-v4-pro

Qwen25

Qwen2.5-72B-Instruct-TurboQwen2.5-7B-Instruct-TurboQwen2.5-Coder-32B-InstructQwen3.5-122B-A10BQwen3.6-27B-FP8Qwen3.6-35B-A3B-FP8qwen-2.5-7b-instructqwen3-32bqwen3-coder-nextqwen3-max-thinkingqwen3.5-122b-a10bqwen3.5-27bqwen3.5-35b-a3bqwen3.5-397b-a17bqwen3.5-9bqwen3.5-flash-02-23qwen3.5-plus-02-15qwen3.5-plus-20260420qwen3.6-27bqwen3.6-35b-a3bqwen3.6-flashqwen3.6-max-previewqwen3.6-plusqwen3.7-maxqwen3.7-plus

Moonshot · Kimi4

kimi-k2.5kimi-k2.6kimi-k2.7-codekimi-latest

Zhipu · GLM7

glm-4.7glm-4.7-flashglm-5glm-5-turboglm-5.1glm-5.2glm-5v-turbo

xAI · Grok4

grok-4.20grok-4.20-multi-agentgrok-4.3grok-build-0.1

Meta · Llama7

Llama-3-3-70B-Instruct-TurboMeta-Llama-3-70B-Instruct-TurboMeta-Llama-3.1-405B-Instruct-TurboMeta-Llama-3.1-70B-Instruct-TurboMeta-Llama-3.1-8B-Instruct-Turbollama-3.1-8b-instructllama-3.3-70b-instruct

Mistral7

Mistral-7B-Instruct-v0.1Mistral-7B-Instruct-v0.3Mixtral-8x22B-Instruct-v0.1Mixtral-8x7B-Instruct-v0.1mistral-medium-3-5mistral-nemomistral-small-2603

MiniMax5

minimax-m2-herminimax-m2.1minimax-m2.5minimax-m2.7minimax-m3

ByteDance · Seed4

seed-1.6seed-1.6-flashseed-2.0-liteseed-2.0-mini

Community11

deepseek-r1:1.5bdeepseek-v4-flashdeepseek-v4-progemma3:4bllama-3.1-8b-instantllama3.1llama3.2llama3.2:1bphi3:latestqwen2.5-coder:3bqwen3:4b

Poolside6

laguna-m.1laguna-m.1:freelaguna-xs-2.1laguna-xs-2.1:freelaguna-xs.2laguna-xs.2:free

Nvidia5

nemotron-3-nano-omni-30b-a3b-reasoning:freenemotron-3-super-120b-a12bnemotron-3-ultra-550b-a55bnemotron-3-ultra-550b-a55b:freenemotron-3.5-content-safety:free

inclusionAI3

ling-2.6-1tling-2.6-flashring-2.6-1t

Cohere2

command-r-08-2024north-mini-code:free

DeepSeek2

DeepSeek-Coder-V2-Lite-Instruct"DeepSeek-V4-Flash

StepFun2

step-3.5-flashstep-3.7-flash

Xiaomi2

mimo-v2.5mimo-v2.5-pro

"meta-llama1

Meta-Llama-3-8B-Instruct-Turbo

AionLabs1

aion-2.0

Amazon1

nova-lite-v1

Arcee1

trinity-large-thinking

Huihui_ai1

qwen3.5-abliterated:27b

IBM Granite1

granite-4.1-8b

Inception1

mercury-2

Kwaipilot1

kat-coder-pro-v2

Liquid1

lfm-2-24b-a2b

Microsoft1

phi-4

Nex-agi1

nex-n2-pro

Nousresearch1

Nous-Hermes-2-Mixtral-8x7B-DPO

Openrouter1

fusion

Perceptron1

perceptron-mk1

Rekaai1

reka-edge

Sakana1

fugu-ultra

Tencent1

hy3-preview

Upstage1

solar-pro-3

Writer1

palmyra-x5

Zhipu · GLM1

GLM-5.1-FP8

Chinese model

DeepSeek

The open Chinese model everyone's talking about — served peer-to-peer on Halo. No account, no subscription, no phone number. Pay only per prompt.

Chinese model

Qwen

Alibaba's powerful open model family — served peer-to-peer on Halo. No account, no subscription, no phone number. Pay only per prompt.

Chinese model

Kimi

Moonshot's Kimi — long-context, strong at reasoning and code — served peer-to-peer on Halo. No account, no subscription. Pay only per prompt.

Chinese model

GLM

Zhipu's GLM family — a capable, efficient open model line — served peer-to-peer on Halo. No account, no subscription. Pay only per prompt.

Chinese model

MiniMax

MiniMax's efficient agentic models — served peer-to-peer on Halo. No account, no subscription. Pay only per prompt.

Overview

Chinese ChatGPT alternatives

The Chinese models worth knowing — DeepSeek, Qwen, Kimi, GLM, MiniMax — and how to run any of them on Halo without a subscription or account.

Comparison

DeepSeek vs ChatGPT

How DeepSeek and ChatGPT stack up on cost, reasoning, privacy and access — and how to run either one on Halo without a subscription.

Access

ChatGPT & Claude in Russia

OpenAI and Anthropic don't serve Russia — but you can still reach GPT-5, Claude and 140+ models on Halo. No account, no subscription, pay per prompt in USDC.

Access

ChatGPT & Claude in China

OpenAI and Anthropic don't serve mainland China — but you can still reach GPT-5, Claude and 140+ models on Halo. No account, no subscription, pay per prompt in USDC.

Access

ChatGPT & Claude in Hong Kong

OpenAI and Anthropic don't offer service in Hong Kong — but you can still reach GPT-5, Claude and 140+ models on Halo. No account, no subscription, pay per prompt in USDC.